Hugging Face's logo

Spaces:
kengboon
/
anomaly-detection
Running

App Files
KB Teo commited on
Commit
22844af
·
1 Parent(s): 508121b

Upload model

Browse files
Files changed (3) hide show
  1. models/glass-tr-5-ov/metadata.json +60 -0
  2. models/glass-tr-5-ov/model.bin +3 -0
  3. models/glass-tr-5-ov/model.xml +2606 -0
models/glass-tr-5-ov/metadata.json ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task": "segmentation",
3
+ "transform": {
4
+ "__version__": "1.3.1",
5
+ "transform": {
6
+ "__class_fullname__": "Compose",
7
+ "p": 1.0,
8
+ "transforms": [
9
+ {
10
+ "__class_fullname__": "Resize",
11
+ "always_apply": true,
12
+ "p": 1,
13
+ "height": 256,
14
+ "width": 341,
15
+ "interpolation": 1
16
+ },
17
+ {
18
+ "__class_fullname__": "CenterCrop",
19
+ "always_apply": true,
20
+ "p": 1.0,
21
+ "height": 256,
22
+ "width": 256
23
+ },
24
+ {
25
+ "__class_fullname__": "Normalize",
26
+ "always_apply": false,
27
+ "p": 1.0,
28
+ "mean": [
29
+ 0.485,
30
+ 0.456,
31
+ 0.406
32
+ ],
33
+ "std": [
34
+ 0.229,
35
+ 0.224,
36
+ 0.225
37
+ ],
38
+ "max_pixel_value": 255.0
39
+ },
40
+ {
41
+ "__class_fullname__": "ToTensorV2",
42
+ "always_apply": true,
43
+ "p": 1.0,
44
+ "transpose_mask": false
45
+ }
46
+ ],
47
+ "bbox_params": null,
48
+ "keypoint_params": null,
49
+ "additional_targets": {
50
+ "image": "image",
51
+ "depth_image": "image"
52
+ },
53
+ "is_check_shapes": true
54
+ }
55
+ },
56
+ "image_threshold": 1.8665753602981567,
57
+ "pixel_threshold": 1.593752145767212,
58
+ "min": 5.206725745665608e-06,
59
+ "max": 33.02467727661133
60
+ }
models/glass-tr-5-ov/model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fed40ef097b6ba12f2ae406e3f90ea40d5e2d6c6893bbb469b612b2a23501021
3
+ size 86750652
models/glass-tr-5-ov/model.xml ADDED
@@ -0,0 +1,2606 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ <?xml version="1.0"?>
2
+ <net name="main_graph" version="11">
3
+ <layers>
4
+ <layer id="0" name="input" type="Parameter" version="opset1">
5
+ <data shape="?,3,256,256" element_type="f32" />
6
+ <rt_info>
7
+ <attribute name="old_api_map_element_type" version="0" value="f16" />
8
+ </rt_info>
9
+ <output>
10
+ <port id="0" precision="FP32" names="input">
11
+ <dim>-1</dim>
12
+ <dim>3</dim>
13
+ <dim>256</dim>
14
+ <dim>256</dim>
15
+ </port>
16
+ </output>
17
+ </layer>
18
+ <layer id="1" name="onnx::Conv_280_compressed" type="Const" version="opset1">
19
+ <data element_type="f16" shape="64, 3, 7, 7" offset="0" size="18816" />
20
+ <output>
21
+ <port id="0" precision="FP16" names="onnx::Conv_280">
22
+ <dim>64</dim>
23
+ <dim>3</dim>
24
+ <dim>7</dim>
25
+ <dim>7</dim>
26
+ </port>
27
+ </output>
28
+ </layer>
29
+ <layer id="2" name="onnx::Conv_280" type="Convert" version="opset1">
30
+ <data destination_type="f32" />
31
+ <rt_info>
32
+ <attribute name="decompression" version="0" />
33
+ </rt_info>
34
+ <input>
35
+ <port id="0" precision="FP16">
36
+ <dim>64</dim>
37
+ <dim>3</dim>
38
+ <dim>7</dim>
39
+ <dim>7</dim>
40
+ </port>
41
+ </input>
42
+ <output>
43
+ <port id="1" precision="FP32">
44
+ <dim>64</dim>
45
+ <dim>3</dim>
46
+ <dim>7</dim>
47
+ <dim>7</dim>
48
+ </port>
49
+ </output>
50
+ </layer>
51
+ <layer id="3" name="/feature_extractor/feature_extractor/conv1/Conv/WithoutBiases" type="Convolution" version="opset1">
52
+ <data strides="2, 2" dilations="1, 1" pads_begin="3, 3" pads_end="3, 3" auto_pad="explicit" />
53
+ <input>
54
+ <port id="0" precision="FP32">
55
+ <dim>-1</dim>
56
+ <dim>3</dim>
57
+ <dim>256</dim>
58
+ <dim>256</dim>
59
+ </port>
60
+ <port id="1" precision="FP32">
61
+ <dim>64</dim>
62
+ <dim>3</dim>
63
+ <dim>7</dim>
64
+ <dim>7</dim>
65
+ </port>
66
+ </input>
67
+ <output>
68
+ <port id="2" precision="FP32" names="/feature_extractor/feature_extractor/conv1/Conv_output_0">
69
+ <dim>-1</dim>
70
+ <dim>64</dim>
71
+ <dim>128</dim>
72
+ <dim>128</dim>
73
+ </port>
74
+ </output>
75
+ </layer>
76
+ <layer id="4" name="/feature_extractor/feature_extractor/act1/Relu" type="ReLU" version="opset1">
77
+ <input>
78
+ <port id="0" precision="FP32">
79
+ <dim>-1</dim>
80
+ <dim>64</dim>
81
+ <dim>128</dim>
82
+ <dim>128</dim>
83
+ </port>
84
+ </input>
85
+ <output>
86
+ <port id="1" precision="FP32" names="/feature_extractor/feature_extractor/act1/Relu_output_0">
87
+ <dim>-1</dim>
88
+ <dim>64</dim>
89
+ <dim>128</dim>
90
+ <dim>128</dim>
91
+ </port>
92
+ </output>
93
+ </layer>
94
+ <layer id="5" name="/feature_extractor/feature_extractor/maxpool/MaxPool" type="MaxPool" version="opset8">
95
+ <data strides="2, 2" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" kernel="3, 3" rounding_type="floor" auto_pad="explicit" index_element_type="i64" axis="0" />
96
+ <input>
97
+ <port id="0" precision="FP32">
98
+ <dim>-1</dim>
99
+ <dim>64</dim>
100
+ <dim>128</dim>
101
+ <dim>128</dim>
102
+ </port>
103
+ </input>
104
+ <output>
105
+ <port id="1" precision="FP32" names="/feature_extractor/feature_extractor/maxpool/MaxPool_output_0">
106
+ <dim>-1</dim>
107
+ <dim>64</dim>
108
+ <dim>64</dim>
109
+ <dim>64</dim>
110
+ </port>
111
+ <port id="2" precision="I64">
112
+ <dim>-1</dim>
113
+ <dim>64</dim>
114
+ <dim>64</dim>
115
+ <dim>64</dim>
116
+ </port>
117
+ </output>
118
+ </layer>
119
+ <layer id="6" name="onnx::Conv_283_compressed" type="Const" version="opset1">
120
+ <data element_type="f16" shape="64, 64, 3, 3" offset="18816" size="73728" />
121
+ <output>
122
+ <port id="0" precision="FP16" names="onnx::Conv_283">
123
+ <dim>64</dim>
124
+ <dim>64</dim>
125
+ <dim>3</dim>
126
+ <dim>3</dim>
127
+ </port>
128
+ </output>
129
+ </layer>
130
+ <layer id="7" name="onnx::Conv_283" type="Convert" version="opset1">
131
+ <data destination_type="f32" />
132
+ <rt_info>
133
+ <attribute name="decompression" version="0" />
134
+ </rt_info>
135
+ <input>
136
+ <port id="0" precision="FP16">
137
+ <dim>64</dim>
138
+ <dim>64</dim>
139
+ <dim>3</dim>
140
+ <dim>3</dim>
141
+ </port>
142
+ </input>
143
+ <output>
144
+ <port id="1" precision="FP32">
145
+ <dim>64</dim>
146
+ <dim>64</dim>
147
+ <dim>3</dim>
148
+ <dim>3</dim>
149
+ </port>
150
+ </output>
151
+ </layer>
152
+ <layer id="8" name="/feature_extractor/feature_extractor/layer1/layer1.0/conv1/Conv/WithoutBiases" type="Convolution" version="opset1">
153
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
154
+ <input>
155
+ <port id="0" precision="FP32">
156
+ <dim>-1</dim>
157
+ <dim>64</dim>
158
+ <dim>64</dim>
159
+ <dim>64</dim>
160
+ </port>
161
+ <port id="1" precision="FP32">
162
+ <dim>64</dim>
163
+ <dim>64</dim>
164
+ <dim>3</dim>
165
+ <dim>3</dim>
166
+ </port>
167
+ </input>
168
+ <output>
169
+ <port id="2" precision="FP32" names="/feature_extractor/feature_extractor/layer1/layer1.0/conv1/Conv_output_0">
170
+ <dim>-1</dim>
171
+ <dim>64</dim>
172
+ <dim>64</dim>
173
+ <dim>64</dim>
174
+ </port>
175
+ </output>
176
+ </layer>
177
+ <layer id="9" name="/feature_extractor/feature_extractor/layer1/layer1.0/act1/Relu" type="ReLU" version="opset1">
178
+ <input>
179
+ <port id="0" precision="FP32">
180
+ <dim>-1</dim>
181
+ <dim>64</dim>
182
+ <dim>64</dim>
183
+ <dim>64</dim>
184
+ </port>
185
+ </input>
186
+ <output>
187
+ <port id="1" precision="FP32" names="/feature_extractor/feature_extractor/layer1/layer1.0/act1/Relu_output_0">
188
+ <dim>-1</dim>
189
+ <dim>64</dim>
190
+ <dim>64</dim>
191
+ <dim>64</dim>
192
+ </port>
193
+ </output>
194
+ </layer>
195
+ <layer id="10" name="onnx::Conv_292_compressed" type="Const" version="opset1">
196
+ <data element_type="f16" shape="64, 64, 3, 3" offset="92544" size="73728" />
197
+ <output>
198
+ <port id="0" precision="FP16" names="onnx::Conv_286,onnx::Conv_292">
199
+ <dim>64</dim>
200
+ <dim>64</dim>
201
+ <dim>3</dim>
202
+ <dim>3</dim>
203
+ </port>
204
+ </output>
205
+ </layer>
206
+ <layer id="11" name="onnx::Conv_292" type="Convert" version="opset1">
207
+ <data destination_type="f32" />
208
+ <rt_info>
209
+ <attribute name="decompression" version="0" />
210
+ </rt_info>
211
+ <input>
212
+ <port id="0" precision="FP16">
213
+ <dim>64</dim>
214
+ <dim>64</dim>
215
+ <dim>3</dim>
216
+ <dim>3</dim>
217
+ </port>
218
+ </input>
219
+ <output>
220
+ <port id="1" precision="FP32">
221
+ <dim>64</dim>
222
+ <dim>64</dim>
223
+ <dim>3</dim>
224
+ <dim>3</dim>
225
+ </port>
226
+ </output>
227
+ </layer>
228
+ <layer id="12" name="/feature_extractor/feature_extractor/layer1/layer1.0/conv2/Conv/WithoutBiases" type="Convolution" version="opset1">
229
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
230
+ <input>
231
+ <port id="0" precision="FP32">
232
+ <dim>-1</dim>
233
+ <dim>64</dim>
234
+ <dim>64</dim>
235
+ <dim>64</dim>
236
+ </port>
237
+ <port id="1" precision="FP32">
238
+ <dim>64</dim>
239
+ <dim>64</dim>
240
+ <dim>3</dim>
241
+ <dim>3</dim>
242
+ </port>
243
+ </input>
244
+ <output>
245
+ <port id="2" precision="FP32" names="/feature_extractor/feature_extractor/layer1/layer1.0/conv2/Conv_output_0">
246
+ <dim>-1</dim>
247
+ <dim>64</dim>
248
+ <dim>64</dim>
249
+ <dim>64</dim>
250
+ </port>
251
+ </output>
252
+ </layer>
253
+ <layer id="13" name="/feature_extractor/feature_extractor/layer1/layer1.0/Add" type="Add" version="opset1">
254
+ <data auto_broadcast="numpy" />
255
+ <input>
256
+ <port id="0" precision="FP32">
257
+ <dim>-1</dim>
258
+ <dim>64</dim>
259
+ <dim>64</dim>
260
+ <dim>64</dim>
261
+ </port>
262
+ <port id="1" precision="FP32">
263
+ <dim>-1</dim>
264
+ <dim>64</dim>
265
+ <dim>64</dim>
266
+ <dim>64</dim>
267
+ </port>
268
+ </input>
269
+ <output>
270
+ <port id="2" precision="FP32" names="/feature_extractor/feature_extractor/layer1/layer1.0/Add_output_0">
271
+ <dim>-1</dim>
272
+ <dim>64</dim>
273
+ <dim>64</dim>
274
+ <dim>64</dim>
275
+ </port>
276
+ </output>
277
+ </layer>
278
+ <layer id="14" name="/feature_extractor/feature_extractor/layer1/layer1.0/act2/Relu" type="ReLU" version="opset1">
279
+ <input>
280
+ <port id="0" precision="FP32">
281
+ <dim>-1</dim>
282
+ <dim>64</dim>
283
+ <dim>64</dim>
284
+ <dim>64</dim>
285
+ </port>
286
+ </input>
287
+ <output>
288
+ <port id="1" precision="FP32" names="/feature_extractor/feature_extractor/layer1/layer1.0/act2/Relu_output_0">
289
+ <dim>-1</dim>
290
+ <dim>64</dim>
291
+ <dim>64</dim>
292
+ <dim>64</dim>
293
+ </port>
294
+ </output>
295
+ </layer>
296
+ <layer id="15" name="onnx::Conv_289_compressed" type="Const" version="opset1">
297
+ <data element_type="f16" shape="64, 64, 3, 3" offset="166272" size="73728" />
298
+ <output>
299
+ <port id="0" precision="FP16" names="onnx::Conv_289">
300
+ <dim>64</dim>
301
+ <dim>64</dim>
302
+ <dim>3</dim>
303
+ <dim>3</dim>
304
+ </port>
305
+ </output>
306
+ </layer>
307
+ <layer id="16" name="onnx::Conv_289" type="Convert" version="opset1">
308
+ <data destination_type="f32" />
309
+ <rt_info>
310
+ <attribute name="decompression" version="0" />
311
+ </rt_info>
312
+ <input>
313
+ <port id="0" precision="FP16">
314
+ <dim>64</dim>
315
+ <dim>64</dim>
316
+ <dim>3</dim>
317
+ <dim>3</dim>
318
+ </port>
319
+ </input>
320
+ <output>
321
+ <port id="1" precision="FP32">
322
+ <dim>64</dim>
323
+ <dim>64</dim>
324
+ <dim>3</dim>
325
+ <dim>3</dim>
326
+ </port>
327
+ </output>
328
+ </layer>
329
+ <layer id="17" name="/feature_extractor/feature_extractor/layer1/layer1.1/conv1/Conv/WithoutBiases" type="Convolution" version="opset1">
330
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
331
+ <input>
332
+ <port id="0" precision="FP32">
333
+ <dim>-1</dim>
334
+ <dim>64</dim>
335
+ <dim>64</dim>
336
+ <dim>64</dim>
337
+ </port>
338
+ <port id="1" precision="FP32">
339
+ <dim>64</dim>
340
+ <dim>64</dim>
341
+ <dim>3</dim>
342
+ <dim>3</dim>
343
+ </port>
344
+ </input>
345
+ <output>
346
+ <port id="2" precision="FP32" names="/feature_extractor/feature_extractor/layer1/layer1.1/conv1/Conv_output_0">
347
+ <dim>-1</dim>
348
+ <dim>64</dim>
349
+ <dim>64</dim>
350
+ <dim>64</dim>
351
+ </port>
352
+ </output>
353
+ </layer>
354
+ <layer id="18" name="/feature_extractor/feature_extractor/layer1/layer1.1/act1/Relu" type="ReLU" version="opset1">
355
+ <input>
356
+ <port id="0" precision="FP32">
357
+ <dim>-1</dim>
358
+ <dim>64</dim>
359
+ <dim>64</dim>
360
+ <dim>64</dim>
361
+ </port>
362
+ </input>
363
+ <output>
364
+ <port id="1" precision="FP32" names="/feature_extractor/feature_extractor/layer1/layer1.1/act1/Relu_output_0">
365
+ <dim>-1</dim>
366
+ <dim>64</dim>
367
+ <dim>64</dim>
368
+ <dim>64</dim>
369
+ </port>
370
+ </output>
371
+ </layer>
372
+ <layer id="19" name="/feature_extractor/feature_extractor/layer1/layer1.1/conv2/Conv/WithoutBiases" type="Convolution" version="opset1">
373
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
374
+ <input>
375
+ <port id="0" precision="FP32">
376
+ <dim>-1</dim>
377
+ <dim>64</dim>
378
+ <dim>64</dim>
379
+ <dim>64</dim>
380
+ </port>
381
+ <port id="1" precision="FP32">
382
+ <dim>64</dim>
383
+ <dim>64</dim>
384
+ <dim>3</dim>
385
+ <dim>3</dim>
386
+ </port>
387
+ </input>
388
+ <output>
389
+ <port id="2" precision="FP32" names="/feature_extractor/feature_extractor/layer1/layer1.1/conv2/Conv_output_0">
390
+ <dim>-1</dim>
391
+ <dim>64</dim>
392
+ <dim>64</dim>
393
+ <dim>64</dim>
394
+ </port>
395
+ </output>
396
+ </layer>
397
+ <layer id="20" name="/feature_extractor/feature_extractor/layer1/layer1.1/Add" type="Add" version="opset1">
398
+ <data auto_broadcast="numpy" />
399
+ <input>
400
+ <port id="0" precision="FP32">
401
+ <dim>-1</dim>
402
+ <dim>64</dim>
403
+ <dim>64</dim>
404
+ <dim>64</dim>
405
+ </port>
406
+ <port id="1" precision="FP32">
407
+ <dim>-1</dim>
408
+ <dim>64</dim>
409
+ <dim>64</dim>
410
+ <dim>64</dim>
411
+ </port>
412
+ </input>
413
+ <output>
414
+ <port id="2" precision="FP32" names="/feature_extractor/feature_extractor/layer1/layer1.1/Add_output_0">
415
+ <dim>-1</dim>
416
+ <dim>64</dim>
417
+ <dim>64</dim>
418
+ <dim>64</dim>
419
+ </port>
420
+ </output>
421
+ </layer>
422
+ <layer id="21" name="/feature_extractor/feature_extractor/layer1/layer1.1/act2/Relu" type="ReLU" version="opset1">
423
+ <input>
424
+ <port id="0" precision="FP32">
425
+ <dim>-1</dim>
426
+ <dim>64</dim>
427
+ <dim>64</dim>
428
+ <dim>64</dim>
429
+ </port>
430
+ </input>
431
+ <output>
432
+ <port id="1" precision="FP32" names="/feature_extractor/feature_extractor/layer1/layer1.1/act2/Relu_output_0">
433
+ <dim>-1</dim>
434
+ <dim>64</dim>
435
+ <dim>64</dim>
436
+ <dim>64</dim>
437
+ </port>
438
+ </output>
439
+ </layer>
440
+ <layer id="22" name="onnx::Conv_295_compressed" type="Const" version="opset1">
441
+ <data element_type="f16" shape="128, 64, 3, 3" offset="240000" size="147456" />
442
+ <output>
443
+ <port id="0" precision="FP16" names="onnx::Conv_295">
444
+ <dim>128</dim>
445
+ <dim>64</dim>
446
+ <dim>3</dim>
447
+ <dim>3</dim>
448
+ </port>
449
+ </output>
450
+ </layer>
451
+ <layer id="23" name="onnx::Conv_295" type="Convert" version="opset1">
452
+ <data destination_type="f32" />
453
+ <rt_info>
454
+ <attribute name="decompression" version="0" />
455
+ </rt_info>
456
+ <input>
457
+ <port id="0" precision="FP16">
458
+ <dim>128</dim>
459
+ <dim>64</dim>
460
+ <dim>3</dim>
461
+ <dim>3</dim>
462
+ </port>
463
+ </input>
464
+ <output>
465
+ <port id="1" precision="FP32">
466
+ <dim>128</dim>
467
+ <dim>64</dim>
468
+ <dim>3</dim>
469
+ <dim>3</dim>
470
+ </port>
471
+ </output>
472
+ </layer>
473
+ <layer id="24" name="/feature_extractor/feature_extractor/layer2/layer2.0/conv1/Conv/WithoutBiases" type="Convolution" version="opset1">
474
+ <data strides="2, 2" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
475
+ <input>
476
+ <port id="0" precision="FP32">
477
+ <dim>-1</dim>
478
+ <dim>64</dim>
479
+ <dim>64</dim>
480
+ <dim>64</dim>
481
+ </port>
482
+ <port id="1" precision="FP32">
483
+ <dim>128</dim>
484
+ <dim>64</dim>
485
+ <dim>3</dim>
486
+ <dim>3</dim>
487
+ </port>
488
+ </input>
489
+ <output>
490
+ <port id="2" precision="FP32" names="/feature_extractor/feature_extractor/layer2/layer2.0/conv1/Conv_output_0">
491
+ <dim>-1</dim>
492
+ <dim>128</dim>
493
+ <dim>32</dim>
494
+ <dim>32</dim>
495
+ </port>
496
+ </output>
497
+ </layer>
498
+ <layer id="25" name="/feature_extractor/feature_extractor/layer2/layer2.0/act1/Relu" type="ReLU" version="opset1">
499
+ <input>
500
+ <port id="0" precision="FP32">
501
+ <dim>-1</dim>
502
+ <dim>128</dim>
503
+ <dim>32</dim>
504
+ <dim>32</dim>
505
+ </port>
506
+ </input>
507
+ <output>
508
+ <port id="1" precision="FP32" names="/feature_extractor/feature_extractor/layer2/layer2.0/act1/Relu_output_0">
509
+ <dim>-1</dim>
510
+ <dim>128</dim>
511
+ <dim>32</dim>
512
+ <dim>32</dim>
513
+ </port>
514
+ </output>
515
+ </layer>
516
+ <layer id="26" name="onnx::Conv_307_compressed" type="Const" version="opset1">
517
+ <data element_type="f16" shape="128, 128, 3, 3" offset="387456" size="294912" />
518
+ <output>
519
+ <port id="0" precision="FP16" names="onnx::Conv_298,onnx::Conv_307">
520
+ <dim>128</dim>
521
+ <dim>128</dim>
522
+ <dim>3</dim>
523
+ <dim>3</dim>
524
+ </port>
525
+ </output>
526
+ </layer>
527
+ <layer id="27" name="onnx::Conv_307" type="Convert" version="opset1">
528
+ <data destination_type="f32" />
529
+ <rt_info>
530
+ <attribute name="decompression" version="0" />
531
+ </rt_info>
532
+ <input>
533
+ <port id="0" precision="FP16">
534
+ <dim>128</dim>
535
+ <dim>128</dim>
536
+ <dim>3</dim>
537
+ <dim>3</dim>
538
+ </port>
539
+ </input>
540
+ <output>
541
+ <port id="1" precision="FP32">
542
+ <dim>128</dim>
543
+ <dim>128</dim>
544
+ <dim>3</dim>
545
+ <dim>3</dim>
546
+ </port>
547
+ </output>
548
+ </layer>
549
+ <layer id="28" name="/feature_extractor/feature_extractor/layer2/layer2.0/conv2/Conv/WithoutBiases" type="Convolution" version="opset1">
550
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
551
+ <input>
552
+ <port id="0" precision="FP32">
553
+ <dim>-1</dim>
554
+ <dim>128</dim>
555
+ <dim>32</dim>
556
+ <dim>32</dim>
557
+ </port>
558
+ <port id="1" precision="FP32">
559
+ <dim>128</dim>
560
+ <dim>128</dim>
561
+ <dim>3</dim>
562
+ <dim>3</dim>
563
+ </port>
564
+ </input>
565
+ <output>
566
+ <port id="2" precision="FP32" names="/feature_extractor/feature_extractor/layer2/layer2.0/conv2/Conv_output_0">
567
+ <dim>-1</dim>
568
+ <dim>128</dim>
569
+ <dim>32</dim>
570
+ <dim>32</dim>
571
+ </port>
572
+ </output>
573
+ </layer>
574
+ <layer id="29" name="onnx::Conv_301_compressed" type="Const" version="opset1">
575
+ <data element_type="f16" shape="128, 64, 1, 1" offset="682368" size="16384" />
576
+ <output>
577
+ <port id="0" precision="FP16" names="onnx::Conv_301">
578
+ <dim>128</dim>
579
+ <dim>64</dim>
580
+ <dim>1</dim>
581
+ <dim>1</dim>
582
+ </port>
583
+ </output>
584
+ </layer>
585
+ <layer id="30" name="onnx::Conv_301" type="Convert" version="opset1">
586
+ <data destination_type="f32" />
587
+ <rt_info>
588
+ <attribute name="decompression" version="0" />
589
+ </rt_info>
590
+ <input>
591
+ <port id="0" precision="FP16">
592
+ <dim>128</dim>
593
+ <dim>64</dim>
594
+ <dim>1</dim>
595
+ <dim>1</dim>
596
+ </port>
597
+ </input>
598
+ <output>
599
+ <port id="1" precision="FP32">
600
+ <dim>128</dim>
601
+ <dim>64</dim>
602
+ <dim>1</dim>
603
+ <dim>1</dim>
604
+ </port>
605
+ </output>
606
+ </layer>
607
+ <layer id="31" name="/feature_extractor/feature_extractor/layer2/layer2.0/downsample/downsample.0/Conv/WithoutBiases" type="Convolution" version="opset1">
608
+ <data strides="2, 2" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" auto_pad="explicit" />
609
+ <input>
610
+ <port id="0" precision="FP32">
611
+ <dim>-1</dim>
612
+ <dim>64</dim>
613
+ <dim>64</dim>
614
+ <dim>64</dim>
615
+ </port>
616
+ <port id="1" precision="FP32">
617
+ <dim>128</dim>
618
+ <dim>64</dim>
619
+ <dim>1</dim>
620
+ <dim>1</dim>
621
+ </port>
622
+ </input>
623
+ <output>
624
+ <port id="2" precision="FP32" names="/feature_extractor/feature_extractor/layer2/layer2.0/downsample/downsample.0/Conv_output_0">
625
+ <dim>-1</dim>
626
+ <dim>128</dim>
627
+ <dim>32</dim>
628
+ <dim>32</dim>
629
+ </port>
630
+ </output>
631
+ </layer>
632
+ <layer id="32" name="/feature_extractor/feature_extractor/layer2/layer2.0/Add" type="Add" version="opset1">
633
+ <data auto_broadcast="numpy" />
634
+ <input>
635
+ <port id="0" precision="FP32">
636
+ <dim>-1</dim>
637
+ <dim>128</dim>
638
+ <dim>32</dim>
639
+ <dim>32</dim>
640
+ </port>
641
+ <port id="1" precision="FP32">
642
+ <dim>-1</dim>
643
+ <dim>128</dim>
644
+ <dim>32</dim>
645
+ <dim>32</dim>
646
+ </port>
647
+ </input>
648
+ <output>
649
+ <port id="2" precision="FP32" names="/feature_extractor/feature_extractor/layer2/layer2.0/Add_output_0">
650
+ <dim>-1</dim>
651
+ <dim>128</dim>
652
+ <dim>32</dim>
653
+ <dim>32</dim>
654
+ </port>
655
+ </output>
656
+ </layer>
657
+ <layer id="33" name="/feature_extractor/feature_extractor/layer2/layer2.0/act2/Relu" type="ReLU" version="opset1">
658
+ <input>
659
+ <port id="0" precision="FP32">
660
+ <dim>-1</dim>
661
+ <dim>128</dim>
662
+ <dim>32</dim>
663
+ <dim>32</dim>
664
+ </port>
665
+ </input>
666
+ <output>
667
+ <port id="1" precision="FP32" names="/feature_extractor/feature_extractor/layer2/layer2.0/act2/Relu_output_0">
668
+ <dim>-1</dim>
669
+ <dim>128</dim>
670
+ <dim>32</dim>
671
+ <dim>32</dim>
672
+ </port>
673
+ </output>
674
+ </layer>
675
+ <layer id="34" name="onnx::Conv_304_compressed" type="Const" version="opset1">
676
+ <data element_type="f16" shape="128, 128, 3, 3" offset="698752" size="294912" />
677
+ <output>
678
+ <port id="0" precision="FP16" names="onnx::Conv_304">
679
+ <dim>128</dim>
680
+ <dim>128</dim>
681
+ <dim>3</dim>
682
+ <dim>3</dim>
683
+ </port>
684
+ </output>
685
+ </layer>
686
+ <layer id="35" name="onnx::Conv_304" type="Convert" version="opset1">
687
+ <data destination_type="f32" />
688
+ <rt_info>
689
+ <attribute name="decompression" version="0" />
690
+ </rt_info>
691
+ <input>
692
+ <port id="0" precision="FP16">
693
+ <dim>128</dim>
694
+ <dim>128</dim>
695
+ <dim>3</dim>
696
+ <dim>3</dim>
697
+ </port>
698
+ </input>
699
+ <output>
700
+ <port id="1" precision="FP32">
701
+ <dim>128</dim>
702
+ <dim>128</dim>
703
+ <dim>3</dim>
704
+ <dim>3</dim>
705
+ </port>
706
+ </output>
707
+ </layer>
708
+ <layer id="36" name="/feature_extractor/feature_extractor/layer2/layer2.1/conv1/Conv/WithoutBiases" type="Convolution" version="opset1">
709
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
710
+ <input>
711
+ <port id="0" precision="FP32">
712
+ <dim>-1</dim>
713
+ <dim>128</dim>
714
+ <dim>32</dim>
715
+ <dim>32</dim>
716
+ </port>
717
+ <port id="1" precision="FP32">
718
+ <dim>128</dim>
719
+ <dim>128</dim>
720
+ <dim>3</dim>
721
+ <dim>3</dim>
722
+ </port>
723
+ </input>
724
+ <output>
725
+ <port id="2" precision="FP32" names="/feature_extractor/feature_extractor/layer2/layer2.1/conv1/Conv_output_0">
726
+ <dim>-1</dim>
727
+ <dim>128</dim>
728
+ <dim>32</dim>
729
+ <dim>32</dim>
730
+ </port>
731
+ </output>
732
+ </layer>
733
+ <layer id="37" name="/feature_extractor/feature_extractor/layer2/layer2.1/act1/Relu" type="ReLU" version="opset1">
734
+ <input>
735
+ <port id="0" precision="FP32">
736
+ <dim>-1</dim>
737
+ <dim>128</dim>
738
+ <dim>32</dim>
739
+ <dim>32</dim>
740
+ </port>
741
+ </input>
742
+ <output>
743
+ <port id="1" precision="FP32" names="/feature_extractor/feature_extractor/layer2/layer2.1/act1/Relu_output_0">
744
+ <dim>-1</dim>
745
+ <dim>128</dim>
746
+ <dim>32</dim>
747
+ <dim>32</dim>
748
+ </port>
749
+ </output>
750
+ </layer>
751
+ <layer id="38" name="/feature_extractor/feature_extractor/layer2/layer2.1/conv2/Conv/WithoutBiases" type="Convolution" version="opset1">
752
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
753
+ <input>
754
+ <port id="0" precision="FP32">
755
+ <dim>-1</dim>
756
+ <dim>128</dim>
757
+ <dim>32</dim>
758
+ <dim>32</dim>
759
+ </port>
760
+ <port id="1" precision="FP32">
761
+ <dim>128</dim>
762
+ <dim>128</dim>
763
+ <dim>3</dim>
764
+ <dim>3</dim>
765
+ </port>
766
+ </input>
767
+ <output>
768
+ <port id="2" precision="FP32" names="/feature_extractor/feature_extractor/layer2/layer2.1/conv2/Conv_output_0">
769
+ <dim>-1</dim>
770
+ <dim>128</dim>
771
+ <dim>32</dim>
772
+ <dim>32</dim>
773
+ </port>
774
+ </output>
775
+ </layer>
776
+ <layer id="39" name="/feature_extractor/feature_extractor/layer2/layer2.1/Add" type="Add" version="opset1">
777
+ <data auto_broadcast="numpy" />
778
+ <input>
779
+ <port id="0" precision="FP32">
780
+ <dim>-1</dim>
781
+ <dim>128</dim>
782
+ <dim>32</dim>
783
+ <dim>32</dim>
784
+ </port>
785
+ <port id="1" precision="FP32">
786
+ <dim>-1</dim>
787
+ <dim>128</dim>
788
+ <dim>32</dim>
789
+ <dim>32</dim>
790
+ </port>
791
+ </input>
792
+ <output>
793
+ <port id="2" precision="FP32" names="/feature_extractor/feature_extractor/layer2/layer2.1/Add_output_0">
794
+ <dim>-1</dim>
795
+ <dim>128</dim>
796
+ <dim>32</dim>
797
+ <dim>32</dim>
798
+ </port>
799
+ </output>
800
+ </layer>
801
+ <layer id="40" name="/feature_extractor/feature_extractor/layer2/layer2.1/act2/Relu" type="ReLU" version="opset1">
802
+ <input>
803
+ <port id="0" precision="FP32">
804
+ <dim>-1</dim>
805
+ <dim>128</dim>
806
+ <dim>32</dim>
807
+ <dim>32</dim>
808
+ </port>
809
+ </input>
810
+ <output>
811
+ <port id="1" precision="FP32" names="/feature_extractor/feature_extractor/layer2/layer2.1/act2/Relu_output_0">
812
+ <dim>-1</dim>
813
+ <dim>128</dim>
814
+ <dim>32</dim>
815
+ <dim>32</dim>
816
+ </port>
817
+ </output>
818
+ </layer>
819
+ <layer id="41" name="/Shape_2" type="ShapeOf" version="opset3">
820
+ <data output_type="i64" />
821
+ <input>
822
+ <port id="0" precision="FP32">
823
+ <dim>-1</dim>
824
+ <dim>128</dim>
825
+ <dim>32</dim>
826
+ <dim>32</dim>
827
+ </port>
828
+ </input>
829
+ <output>
830
+ <port id="1" precision="I64" names="/Shape_2_output_0">
831
+ <dim>4</dim>
832
+ </port>
833
+ </output>
834
+ </layer>
835
+ <layer id="42" name="/Constant_3" type="Const" version="opset1">
836
+ <data element_type="i64" shape="1" offset="993664" size="8" />
837
+ <rt_info>
838
+ <attribute name="precise" version="0" />
839
+ </rt_info>
840
+ <output>
841
+ <port id="0" precision="I64" names="/Constant_3_output_0">
842
+ <dim>1</dim>
843
+ </port>
844
+ </output>
845
+ </layer>
846
+ <layer id="43" name="/Constant_4" type="Const" version="opset1">
847
+ <data element_type="i64" shape="1" offset="993672" size="8" />
848
+ <rt_info>
849
+ <attribute name="precise" version="0" />
850
+ </rt_info>
851
+ <output>
852
+ <port id="0" precision="I64" names="/Constant_4_output_0">
853
+ <dim>1</dim>
854
+ </port>
855
+ </output>
856
+ </layer>
857
+ <layer id="44" name="Broadcast_300" type="Const" version="opset1">
858
+ <data element_type="i64" shape="1" offset="993680" size="8" />
859
+ <rt_info>
860
+ <attribute name="precise" version="0" />
861
+ </rt_info>
862
+ <output>
863
+ <port id="0" precision="I64">
864
+ <dim>1</dim>
865
+ </port>
866
+ </output>
867
+ </layer>
868
+ <layer id="45" name="/Slice" type="StridedSlice" version="opset1">
869
+ <data begin_mask="0" end_mask="0" new_axis_mask="" shrink_axis_mask="" ellipsis_mask="" />
870
+ <input>
871
+ <port id="0" precision="I64">
872
+ <dim>4</dim>
873
+ </port>
874
+ <port id="1" precision="I64">
875
+ <dim>1</dim>
876
+ </port>
877
+ <port id="2" precision="I64">
878
+ <dim>1</dim>
879
+ </port>
880
+ <port id="3" precision="I64">
881
+ <dim>1</dim>
882
+ </port>
883
+ </input>
884
+ <output>
885
+ <port id="4" precision="I64" names="/Slice_output_0">
886
+ <dim>2</dim>
887
+ </port>
888
+ </output>
889
+ </layer>
890
+ <layer id="46" name="/Shape" type="ShapeOf" version="opset3">
891
+ <data output_type="i64" />
892
+ <input>
893
+ <port id="0" precision="FP32">
894
+ <dim>-1</dim>
895
+ <dim>64</dim>
896
+ <dim>64</dim>
897
+ <dim>64</dim>
898
+ </port>
899
+ </input>
900
+ <output>
901
+ <port id="1" precision="I64" names="/Shape_1_output_0,/Shape_output_0">
902
+ <dim>4</dim>
903
+ </port>
904
+ </output>
905
+ </layer>
906
+ <layer id="47" name="Constant_2366" type="Const" version="opset1">
907
+ <data element_type="i64" shape="2" offset="993688" size="16" />
908
+ <rt_info>
909
+ <attribute name="precise" version="0" />
910
+ </rt_info>
911
+ <output>
912
+ <port id="0" precision="I64">
913
+ <dim>2</dim>
914
+ </port>
915
+ </output>
916
+ </layer>
917
+ <layer id="48" name="Constant_2367" type="Const" version="opset1">
918
+ <data element_type="i64" shape="" offset="993664" size="8" />
919
+ <rt_info>
920
+ <attribute name="precise" version="0" />
921
+ </rt_info>
922
+ <output>
923
+ <port id="0" precision="I64" />
924
+ </output>
925
+ </layer>
926
+ <layer id="49" name="Gather_2368" type="Gather" version="opset8">
927
+ <data batch_dims="0" />
928
+ <input>
929
+ <port id="0" precision="I64">
930
+ <dim>4</dim>
931
+ </port>
932
+ <port id="1" precision="I64">
933
+ <dim>2</dim>
934
+ </port>
935
+ <port id="2" precision="I64" />
936
+ </input>
937
+ <output>
938
+ <port id="3" precision="I64" names="/Cast_output_0,/Concat_output_0">
939
+ <dim>2</dim>
940
+ </port>
941
+ </output>
942
+ </layer>
943
+ <layer id="50" name="/Concat_1" type="Concat" version="opset1">
944
+ <data axis="0" />
945
+ <input>
946
+ <port id="0" precision="I64">
947
+ <dim>2</dim>
948
+ </port>
949
+ <port id="1" precision="I64">
950
+ <dim>2</dim>
951
+ </port>
952
+ </input>
953
+ <output>
954
+ <port id="2" precision="I64" names="/Concat_1_output_0">
955
+ <dim>4</dim>
956
+ </port>
957
+ </output>
958
+ </layer>
959
+ <layer id="51" name="/Resize" type="Interpolate" version="opset11">
960
+ <data mode="nearest" shape_calculation_mode="sizes" coordinate_transformation_mode="asymmetric" nearest_mode="floor" antialias="false" pads_begin="0, 0, 0, 0" pads_end="0, 0, 0, 0" cube_coeff="-0.75" />
961
+ <input>
962
+ <port id="0" precision="FP32">
963
+ <dim>-1</dim>
964
+ <dim>128</dim>
965
+ <dim>32</dim>
966
+ <dim>32</dim>
967
+ </port>
968
+ <port id="1" precision="I64">
969
+ <dim>4</dim>
970
+ </port>
971
+ </input>
972
+ <output>
973
+ <port id="2" precision="FP32" names="/Resize_output_0">
974
+ <dim>-1</dim>
975
+ <dim>128</dim>
976
+ <dim>64</dim>
977
+ <dim>64</dim>
978
+ </port>
979
+ </output>
980
+ </layer>
981
+ <layer id="52" name="/Concat_2" type="Concat" version="opset1">
982
+ <data axis="1" />
983
+ <input>
984
+ <port id="0" precision="FP32">
985
+ <dim>-1</dim>
986
+ <dim>64</dim>
987
+ <dim>64</dim>
988
+ <dim>64</dim>
989
+ </port>
990
+ <port id="1" precision="FP32">
991
+ <dim>-1</dim>
992
+ <dim>128</dim>
993
+ <dim>64</dim>
994
+ <dim>64</dim>
995
+ </port>
996
+ </input>
997
+ <output>
998
+ <port id="2" precision="FP32" names="/Concat_2_output_0">
999
+ <dim>-1</dim>
1000
+ <dim>192</dim>
1001
+ <dim>64</dim>
1002
+ <dim>64</dim>
1003
+ </port>
1004
+ </output>
1005
+ </layer>
1006
+ <layer id="53" name="onnx::Conv_310_compressed" type="Const" version="opset1">
1007
+ <data element_type="f16" shape="256, 128, 3, 3" offset="993704" size="589824" />
1008
+ <output>
1009
+ <port id="0" precision="FP16" names="onnx::Conv_310">
1010
+ <dim>256</dim>
1011
+ <dim>128</dim>
1012
+ <dim>3</dim>
1013
+ <dim>3</dim>
1014
+ </port>
1015
+ </output>
1016
+ </layer>
1017
+ <layer id="54" name="onnx::Conv_310" type="Convert" version="opset1">
1018
+ <data destination_type="f32" />
1019
+ <rt_info>
1020
+ <attribute name="decompression" version="0" />
1021
+ </rt_info>
1022
+ <input>
1023
+ <port id="0" precision="FP16">
1024
+ <dim>256</dim>
1025
+ <dim>128</dim>
1026
+ <dim>3</dim>
1027
+ <dim>3</dim>
1028
+ </port>
1029
+ </input>
1030
+ <output>
1031
+ <port id="1" precision="FP32">
1032
+ <dim>256</dim>
1033
+ <dim>128</dim>
1034
+ <dim>3</dim>
1035
+ <dim>3</dim>
1036
+ </port>
1037
+ </output>
1038
+ </layer>
1039
+ <layer id="55" name="/feature_extractor/feature_extractor/layer3/layer3.0/conv1/Conv/WithoutBiases" type="Convolution" version="opset1">
1040
+ <data strides="2, 2" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
1041
+ <input>
1042
+ <port id="0" precision="FP32">
1043
+ <dim>-1</dim>
1044
+ <dim>128</dim>
1045
+ <dim>32</dim>
1046
+ <dim>32</dim>
1047
+ </port>
1048
+ <port id="1" precision="FP32">
1049
+ <dim>256</dim>
1050
+ <dim>128</dim>
1051
+ <dim>3</dim>
1052
+ <dim>3</dim>
1053
+ </port>
1054
+ </input>
1055
+ <output>
1056
+ <port id="2" precision="FP32" names="/feature_extractor/feature_extractor/layer3/layer3.0/conv1/Conv_output_0">
1057
+ <dim>-1</dim>
1058
+ <dim>256</dim>
1059
+ <dim>16</dim>
1060
+ <dim>16</dim>
1061
+ </port>
1062
+ </output>
1063
+ </layer>
1064
+ <layer id="56" name="/feature_extractor/feature_extractor/layer3/layer3.0/act1/Relu" type="ReLU" version="opset1">
1065
+ <input>
1066
+ <port id="0" precision="FP32">
1067
+ <dim>-1</dim>
1068
+ <dim>256</dim>
1069
+ <dim>16</dim>
1070
+ <dim>16</dim>
1071
+ </port>
1072
+ </input>
1073
+ <output>
1074
+ <port id="1" precision="FP32" names="/feature_extractor/feature_extractor/layer3/layer3.0/act1/Relu_output_0">
1075
+ <dim>-1</dim>
1076
+ <dim>256</dim>
1077
+ <dim>16</dim>
1078
+ <dim>16</dim>
1079
+ </port>
1080
+ </output>
1081
+ </layer>
1082
+ <layer id="57" name="onnx::Conv_322_compressed" type="Const" version="opset1">
1083
+ <data element_type="f16" shape="256, 256, 3, 3" offset="1583528" size="1179648" />
1084
+ <output>
1085
+ <port id="0" precision="FP16" names="onnx::Conv_313,onnx::Conv_322">
1086
+ <dim>256</dim>
1087
+ <dim>256</dim>
1088
+ <dim>3</dim>
1089
+ <dim>3</dim>
1090
+ </port>
1091
+ </output>
1092
+ </layer>
1093
+ <layer id="58" name="onnx::Conv_322" type="Convert" version="opset1">
1094
+ <data destination_type="f32" />
1095
+ <rt_info>
1096
+ <attribute name="decompression" version="0" />
1097
+ </rt_info>
1098
+ <input>
1099
+ <port id="0" precision="FP16">
1100
+ <dim>256</dim>
1101
+ <dim>256</dim>
1102
+ <dim>3</dim>
1103
+ <dim>3</dim>
1104
+ </port>
1105
+ </input>
1106
+ <output>
1107
+ <port id="1" precision="FP32">
1108
+ <dim>256</dim>
1109
+ <dim>256</dim>
1110
+ <dim>3</dim>
1111
+ <dim>3</dim>
1112
+ </port>
1113
+ </output>
1114
+ </layer>
1115
+ <layer id="59" name="/feature_extractor/feature_extractor/layer3/layer3.0/conv2/Conv/WithoutBiases" type="Convolution" version="opset1">
1116
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
1117
+ <input>
1118
+ <port id="0" precision="FP32">
1119
+ <dim>-1</dim>
1120
+ <dim>256</dim>
1121
+ <dim>16</dim>
1122
+ <dim>16</dim>
1123
+ </port>
1124
+ <port id="1" precision="FP32">
1125
+ <dim>256</dim>
1126
+ <dim>256</dim>
1127
+ <dim>3</dim>
1128
+ <dim>3</dim>
1129
+ </port>
1130
+ </input>
1131
+ <output>
1132
+ <port id="2" precision="FP32" names="/feature_extractor/feature_extractor/layer3/layer3.0/conv2/Conv_output_0">
1133
+ <dim>-1</dim>
1134
+ <dim>256</dim>
1135
+ <dim>16</dim>
1136
+ <dim>16</dim>
1137
+ </port>
1138
+ </output>
1139
+ </layer>
1140
+ <layer id="60" name="onnx::Conv_316_compressed" type="Const" version="opset1">
1141
+ <data element_type="f16" shape="256, 128, 1, 1" offset="2763176" size="65536" />
1142
+ <output>
1143
+ <port id="0" precision="FP16" names="onnx::Conv_316">
1144
+ <dim>256</dim>
1145
+ <dim>128</dim>
1146
+ <dim>1</dim>
1147
+ <dim>1</dim>
1148
+ </port>
1149
+ </output>
1150
+ </layer>
1151
+ <layer id="61" name="onnx::Conv_316" type="Convert" version="opset1">
1152
+ <data destination_type="f32" />
1153
+ <rt_info>
1154
+ <attribute name="decompression" version="0" />
1155
+ </rt_info>
1156
+ <input>
1157
+ <port id="0" precision="FP16">
1158
+ <dim>256</dim>
1159
+ <dim>128</dim>
1160
+ <dim>1</dim>
1161
+ <dim>1</dim>
1162
+ </port>
1163
+ </input>
1164
+ <output>
1165
+ <port id="1" precision="FP32">
1166
+ <dim>256</dim>
1167
+ <dim>128</dim>
1168
+ <dim>1</dim>
1169
+ <dim>1</dim>
1170
+ </port>
1171
+ </output>
1172
+ </layer>
1173
+ <layer id="62" name="/feature_extractor/feature_extractor/layer3/layer3.0/downsample/downsample.0/Conv/WithoutBiases" type="Convolution" version="opset1">
1174
+ <data strides="2, 2" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" auto_pad="explicit" />
1175
+ <input>
1176
+ <port id="0" precision="FP32">
1177
+ <dim>-1</dim>
1178
+ <dim>128</dim>
1179
+ <dim>32</dim>
1180
+ <dim>32</dim>
1181
+ </port>
1182
+ <port id="1" precision="FP32">
1183
+ <dim>256</dim>
1184
+ <dim>128</dim>
1185
+ <dim>1</dim>
1186
+ <dim>1</dim>
1187
+ </port>
1188
+ </input>
1189
+ <output>
1190
+ <port id="2" precision="FP32" names="/feature_extractor/feature_extractor/layer3/layer3.0/downsample/downsample.0/Conv_output_0">
1191
+ <dim>-1</dim>
1192
+ <dim>256</dim>
1193
+ <dim>16</dim>
1194
+ <dim>16</dim>
1195
+ </port>
1196
+ </output>
1197
+ </layer>
1198
+ <layer id="63" name="/feature_extractor/feature_extractor/layer3/layer3.0/Add" type="Add" version="opset1">
1199
+ <data auto_broadcast="numpy" />
1200
+ <input>
1201
+ <port id="0" precision="FP32">
1202
+ <dim>-1</dim>
1203
+ <dim>256</dim>
1204
+ <dim>16</dim>
1205
+ <dim>16</dim>
1206
+ </port>
1207
+ <port id="1" precision="FP32">
1208
+ <dim>-1</dim>
1209
+ <dim>256</dim>
1210
+ <dim>16</dim>
1211
+ <dim>16</dim>
1212
+ </port>
1213
+ </input>
1214
+ <output>
1215
+ <port id="2" precision="FP32" names="/feature_extractor/feature_extractor/layer3/layer3.0/Add_output_0">
1216
+ <dim>-1</dim>
1217
+ <dim>256</dim>
1218
+ <dim>16</dim>
1219
+ <dim>16</dim>
1220
+ </port>
1221
+ </output>
1222
+ </layer>
1223
+ <layer id="64" name="/feature_extractor/feature_extractor/layer3/layer3.0/act2/Relu" type="ReLU" version="opset1">
1224
+ <input>
1225
+ <port id="0" precision="FP32">
1226
+ <dim>-1</dim>
1227
+ <dim>256</dim>
1228
+ <dim>16</dim>
1229
+ <dim>16</dim>
1230
+ </port>
1231
+ </input>
1232
+ <output>
1233
+ <port id="1" precision="FP32" names="/feature_extractor/feature_extractor/layer3/layer3.0/act2/Relu_output_0">
1234
+ <dim>-1</dim>
1235
+ <dim>256</dim>
1236
+ <dim>16</dim>
1237
+ <dim>16</dim>
1238
+ </port>
1239
+ </output>
1240
+ </layer>
1241
+ <layer id="65" name="onnx::Conv_319_compressed" type="Const" version="opset1">
1242
+ <data element_type="f16" shape="256, 256, 3, 3" offset="2828712" size="1179648" />
1243
+ <output>
1244
+ <port id="0" precision="FP16" names="onnx::Conv_319">
1245
+ <dim>256</dim>
1246
+ <dim>256</dim>
1247
+ <dim>3</dim>
1248
+ <dim>3</dim>
1249
+ </port>
1250
+ </output>
1251
+ </layer>
1252
+ <layer id="66" name="onnx::Conv_319" type="Convert" version="opset1">
1253
+ <data destination_type="f32" />
1254
+ <rt_info>
1255
+ <attribute name="decompression" version="0" />
1256
+ </rt_info>
1257
+ <input>
1258
+ <port id="0" precision="FP16">
1259
+ <dim>256</dim>
1260
+ <dim>256</dim>
1261
+ <dim>3</dim>
1262
+ <dim>3</dim>
1263
+ </port>
1264
+ </input>
1265
+ <output>
1266
+ <port id="1" precision="FP32">
1267
+ <dim>256</dim>
1268
+ <dim>256</dim>
1269
+ <dim>3</dim>
1270
+ <dim>3</dim>
1271
+ </port>
1272
+ </output>
1273
+ </layer>
1274
+ <layer id="67" name="/feature_extractor/feature_extractor/layer3/layer3.1/conv1/Conv/WithoutBiases" type="Convolution" version="opset1">
1275
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
1276
+ <input>
1277
+ <port id="0" precision="FP32">
1278
+ <dim>-1</dim>
1279
+ <dim>256</dim>
1280
+ <dim>16</dim>
1281
+ <dim>16</dim>
1282
+ </port>
1283
+ <port id="1" precision="FP32">
1284
+ <dim>256</dim>
1285
+ <dim>256</dim>
1286
+ <dim>3</dim>
1287
+ <dim>3</dim>
1288
+ </port>
1289
+ </input>
1290
+ <output>
1291
+ <port id="2" precision="FP32" names="/feature_extractor/feature_extractor/layer3/layer3.1/conv1/Conv_output_0">
1292
+ <dim>-1</dim>
1293
+ <dim>256</dim>
1294
+ <dim>16</dim>
1295
+ <dim>16</dim>
1296
+ </port>
1297
+ </output>
1298
+ </layer>
1299
+ <layer id="68" name="/feature_extractor/feature_extractor/layer3/layer3.1/act1/Relu" type="ReLU" version="opset1">
1300
+ <input>
1301
+ <port id="0" precision="FP32">
1302
+ <dim>-1</dim>
1303
+ <dim>256</dim>
1304
+ <dim>16</dim>
1305
+ <dim>16</dim>
1306
+ </port>
1307
+ </input>
1308
+ <output>
1309
+ <port id="1" precision="FP32" names="/feature_extractor/feature_extractor/layer3/layer3.1/act1/Relu_output_0">
1310
+ <dim>-1</dim>
1311
+ <dim>256</dim>
1312
+ <dim>16</dim>
1313
+ <dim>16</dim>
1314
+ </port>
1315
+ </output>
1316
+ </layer>
1317
+ <layer id="69" name="/feature_extractor/feature_extractor/layer3/layer3.1/conv2/Conv/WithoutBiases" type="Convolution" version="opset1">
1318
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
1319
+ <input>
1320
+ <port id="0" precision="FP32">
1321
+ <dim>-1</dim>
1322
+ <dim>256</dim>
1323
+ <dim>16</dim>
1324
+ <dim>16</dim>
1325
+ </port>
1326
+ <port id="1" precision="FP32">
1327
+ <dim>256</dim>
1328
+ <dim>256</dim>
1329
+ <dim>3</dim>
1330
+ <dim>3</dim>
1331
+ </port>
1332
+ </input>
1333
+ <output>
1334
+ <port id="2" precision="FP32" names="/feature_extractor/feature_extractor/layer3/layer3.1/conv2/Conv_output_0">
1335
+ <dim>-1</dim>
1336
+ <dim>256</dim>
1337
+ <dim>16</dim>
1338
+ <dim>16</dim>
1339
+ </port>
1340
+ </output>
1341
+ </layer>
1342
+ <layer id="70" name="/feature_extractor/feature_extractor/layer3/layer3.1/Add" type="Add" version="opset1">
1343
+ <data auto_broadcast="numpy" />
1344
+ <input>
1345
+ <port id="0" precision="FP32">
1346
+ <dim>-1</dim>
1347
+ <dim>256</dim>
1348
+ <dim>16</dim>
1349
+ <dim>16</dim>
1350
+ </port>
1351
+ <port id="1" precision="FP32">
1352
+ <dim>-1</dim>
1353
+ <dim>256</dim>
1354
+ <dim>16</dim>
1355
+ <dim>16</dim>
1356
+ </port>
1357
+ </input>
1358
+ <output>
1359
+ <port id="2" precision="FP32" names="/feature_extractor/feature_extractor/layer3/layer3.1/Add_output_0">
1360
+ <dim>-1</dim>
1361
+ <dim>256</dim>
1362
+ <dim>16</dim>
1363
+ <dim>16</dim>
1364
+ </port>
1365
+ </output>
1366
+ </layer>
1367
+ <layer id="71" name="/feature_extractor/feature_extractor/layer3/layer3.1/act2/Relu" type="ReLU" version="opset1">
1368
+ <input>
1369
+ <port id="0" precision="FP32">
1370
+ <dim>-1</dim>
1371
+ <dim>256</dim>
1372
+ <dim>16</dim>
1373
+ <dim>16</dim>
1374
+ </port>
1375
+ </input>
1376
+ <output>
1377
+ <port id="1" precision="FP32" names="/feature_extractor/feature_extractor/layer3/layer3.1/act2/Relu_output_0">
1378
+ <dim>-1</dim>
1379
+ <dim>256</dim>
1380
+ <dim>16</dim>
1381
+ <dim>16</dim>
1382
+ </port>
1383
+ </output>
1384
+ </layer>
1385
+ <layer id="72" name="/Shape_5" type="ShapeOf" version="opset3">
1386
+ <data output_type="i64" />
1387
+ <input>
1388
+ <port id="0" precision="FP32">
1389
+ <dim>-1</dim>
1390
+ <dim>256</dim>
1391
+ <dim>16</dim>
1392
+ <dim>16</dim>
1393
+ </port>
1394
+ </input>
1395
+ <output>
1396
+ <port id="1" precision="I64" names="/Shape_5_output_0">
1397
+ <dim>4</dim>
1398
+ </port>
1399
+ </output>
1400
+ </layer>
1401
+ <layer id="73" name="/Constant_10" type="Const" version="opset1">
1402
+ <data element_type="i64" shape="1" offset="993664" size="8" />
1403
+ <rt_info>
1404
+ <attribute name="precise" version="0" />
1405
+ </rt_info>
1406
+ <output>
1407
+ <port id="0" precision="I64" names="/Constant_10_output_0">
1408
+ <dim>1</dim>
1409
+ </port>
1410
+ </output>
1411
+ </layer>
1412
+ <layer id="74" name="/Constant_11" type="Const" version="opset1">
1413
+ <data element_type="i64" shape="1" offset="993672" size="8" />
1414
+ <rt_info>
1415
+ <attribute name="precise" version="0" />
1416
+ </rt_info>
1417
+ <output>
1418
+ <port id="0" precision="I64" names="/Constant_11_output_0">
1419
+ <dim>1</dim>
1420
+ </port>
1421
+ </output>
1422
+ </layer>
1423
+ <layer id="75" name="Broadcast_342" type="Const" version="opset1">
1424
+ <data element_type="i64" shape="1" offset="993680" size="8" />
1425
+ <rt_info>
1426
+ <attribute name="precise" version="0" />
1427
+ </rt_info>
1428
+ <output>
1429
+ <port id="0" precision="I64">
1430
+ <dim>1</dim>
1431
+ </port>
1432
+ </output>
1433
+ </layer>
1434
+ <layer id="76" name="/Slice_1" type="StridedSlice" version="opset1">
1435
+ <data begin_mask="0" end_mask="0" new_axis_mask="" shrink_axis_mask="" ellipsis_mask="" />
1436
+ <input>
1437
+ <port id="0" precision="I64">
1438
+ <dim>4</dim>
1439
+ </port>
1440
+ <port id="1" precision="I64">
1441
+ <dim>1</dim>
1442
+ </port>
1443
+ <port id="2" precision="I64">
1444
+ <dim>1</dim>
1445
+ </port>
1446
+ <port id="3" precision="I64">
1447
+ <dim>1</dim>
1448
+ </port>
1449
+ </input>
1450
+ <output>
1451
+ <port id="4" precision="I64" names="/Slice_1_output_0">
1452
+ <dim>2</dim>
1453
+ </port>
1454
+ </output>
1455
+ </layer>
1456
+ <layer id="77" name="/Shape_3" type="ShapeOf" version="opset3">
1457
+ <data output_type="i64" />
1458
+ <input>
1459
+ <port id="0" precision="FP32">
1460
+ <dim>-1</dim>
1461
+ <dim>192</dim>
1462
+ <dim>64</dim>
1463
+ <dim>64</dim>
1464
+ </port>
1465
+ </input>
1466
+ <output>
1467
+ <port id="1" precision="I64" names="/Shape_3_output_0,/Shape_4_output_0">
1468
+ <dim>4</dim>
1469
+ </port>
1470
+ </output>
1471
+ </layer>
1472
+ <layer id="78" name="Constant_2370" type="Const" version="opset1">
1473
+ <data element_type="i64" shape="2" offset="993688" size="16" />
1474
+ <rt_info>
1475
+ <attribute name="precise" version="0" />
1476
+ </rt_info>
1477
+ <output>
1478
+ <port id="0" precision="I64">
1479
+ <dim>2</dim>
1480
+ </port>
1481
+ </output>
1482
+ </layer>
1483
+ <layer id="79" name="Constant_2371" type="Const" version="opset1">
1484
+ <data element_type="i64" shape="" offset="993664" size="8" />
1485
+ <rt_info>
1486
+ <attribute name="precise" version="0" />
1487
+ </rt_info>
1488
+ <output>
1489
+ <port id="0" precision="I64" />
1490
+ </output>
1491
+ </layer>
1492
+ <layer id="80" name="Gather_2372" type="Gather" version="opset8">
1493
+ <data batch_dims="0" />
1494
+ <input>
1495
+ <port id="0" precision="I64">
1496
+ <dim>4</dim>
1497
+ </port>
1498
+ <port id="1" precision="I64">
1499
+ <dim>2</dim>
1500
+ </port>
1501
+ <port id="2" precision="I64" />
1502
+ </input>
1503
+ <output>
1504
+ <port id="3" precision="I64" names="/Cast_1_output_0,/Concat_3_output_0">
1505
+ <dim>2</dim>
1506
+ </port>
1507
+ </output>
1508
+ </layer>
1509
+ <layer id="81" name="/Concat_4" type="Concat" version="opset1">
1510
+ <data axis="0" />
1511
+ <input>
1512
+ <port id="0" precision="I64">
1513
+ <dim>2</dim>
1514
+ </port>
1515
+ <port id="1" precision="I64">
1516
+ <dim>2</dim>
1517
+ </port>
1518
+ </input>
1519
+ <output>
1520
+ <port id="2" precision="I64" names="/Concat_4_output_0">
1521
+ <dim>4</dim>
1522
+ </port>
1523
+ </output>
1524
+ </layer>
1525
+ <layer id="82" name="/Resize_1" type="Interpolate" version="opset11">
1526
+ <data mode="nearest" shape_calculation_mode="sizes" coordinate_transformation_mode="asymmetric" nearest_mode="floor" antialias="false" pads_begin="0, 0, 0, 0" pads_end="0, 0, 0, 0" cube_coeff="-0.75" />
1527
+ <input>
1528
+ <port id="0" precision="FP32">
1529
+ <dim>-1</dim>
1530
+ <dim>256</dim>
1531
+ <dim>16</dim>
1532
+ <dim>16</dim>
1533
+ </port>
1534
+ <port id="1" precision="I64">
1535
+ <dim>4</dim>
1536
+ </port>
1537
+ </input>
1538
+ <output>
1539
+ <port id="2" precision="FP32" names="/Resize_1_output_0">
1540
+ <dim>-1</dim>
1541
+ <dim>256</dim>
1542
+ <dim>64</dim>
1543
+ <dim>64</dim>
1544
+ </port>
1545
+ </output>
1546
+ </layer>
1547
+ <layer id="83" name="/Concat_5" type="Concat" version="opset1">
1548
+ <data axis="1" />
1549
+ <input>
1550
+ <port id="0" precision="FP32">
1551
+ <dim>-1</dim>
1552
+ <dim>192</dim>
1553
+ <dim>64</dim>
1554
+ <dim>64</dim>
1555
+ </port>
1556
+ <port id="1" precision="FP32">
1557
+ <dim>-1</dim>
1558
+ <dim>256</dim>
1559
+ <dim>64</dim>
1560
+ <dim>64</dim>
1561
+ </port>
1562
+ </input>
1563
+ <output>
1564
+ <port id="2" precision="FP32" names="/Concat_5_output_0">
1565
+ <dim>-1</dim>
1566
+ <dim>448</dim>
1567
+ <dim>64</dim>
1568
+ <dim>64</dim>
1569
+ </port>
1570
+ </output>
1571
+ </layer>
1572
+ <layer id="84" name="onnx::Gather_324" type="Const" version="opset1">
1573
+ <data element_type="i64" shape="100" offset="4008360" size="800" />
1574
+ <output>
1575
+ <port id="0" precision="I64" names="onnx::Gather_324">
1576
+ <dim>100</dim>
1577
+ </port>
1578
+ </output>
1579
+ </layer>
1580
+ <layer id="85" name="Constant_365" type="Const" version="opset1">
1581
+ <data element_type="i64" shape="" offset="993680" size="8" />
1582
+ <output>
1583
+ <port id="0" precision="I64" />
1584
+ </output>
1585
+ </layer>
1586
+ <layer id="86" name="/Gather_4" type="Gather" version="opset8">
1587
+ <data batch_dims="0" />
1588
+ <input>
1589
+ <port id="0" precision="FP32">
1590
+ <dim>-1</dim>
1591
+ <dim>448</dim>
1592
+ <dim>64</dim>
1593
+ <dim>64</dim>
1594
+ </port>
1595
+ <port id="1" precision="I64">
1596
+ <dim>100</dim>
1597
+ </port>
1598
+ <port id="2" precision="I64" />
1599
+ </input>
1600
+ <output>
1601
+ <port id="3" precision="FP32" names="/Gather_4_output_0">
1602
+ <dim>-1</dim>
1603
+ <dim>100</dim>
1604
+ <dim>64</dim>
1605
+ <dim>64</dim>
1606
+ </port>
1607
+ </output>
1608
+ </layer>
1609
+ <layer id="87" name="Constant_2469" type="Const" version="opset1">
1610
+ <data element_type="i64" shape="2" offset="4009160" size="16" />
1611
+ <rt_info>
1612
+ <attribute name="precise" version="0" />
1613
+ </rt_info>
1614
+ <output>
1615
+ <port id="0" precision="I64">
1616
+ <dim>2</dim>
1617
+ </port>
1618
+ </output>
1619
+ </layer>
1620
+ <layer id="88" name="/anomaly_map_generator/Shape" type="ShapeOf" version="opset3">
1621
+ <data output_type="i64" />
1622
+ <input>
1623
+ <port id="0" precision="FP32">
1624
+ <dim>-1</dim>
1625
+ <dim>100</dim>
1626
+ <dim>64</dim>
1627
+ <dim>64</dim>
1628
+ </port>
1629
+ </input>
1630
+ <output>
1631
+ <port id="1" precision="I64" names="/anomaly_map_generator/Shape_1_output_0,/anomaly_map_generator/Shape_2_output_0,/anomaly_map_generator/Shape_3_output_0,/anomaly_map_generator/Shape_output_0">
1632
+ <dim>4</dim>
1633
+ </port>
1634
+ </output>
1635
+ </layer>
1636
+ <layer id="89" name="Constant_2349" type="Const" version="opset1">
1637
+ <data element_type="i64" shape="1" offset="993672" size="8" />
1638
+ <rt_info>
1639
+ <attribute name="precise" version="0" />
1640
+ </rt_info>
1641
+ <output>
1642
+ <port id="0" precision="I64">
1643
+ <dim>1</dim>
1644
+ </port>
1645
+ </output>
1646
+ </layer>
1647
+ <layer id="90" name="Constant_377" type="Const" version="opset1">
1648
+ <data element_type="i64" shape="" offset="993664" size="8" />
1649
+ <rt_info>
1650
+ <attribute name="precise" version="0" />
1651
+ </rt_info>
1652
+ <output>
1653
+ <port id="0" precision="I64" />
1654
+ </output>
1655
+ </layer>
1656
+ <layer id="91" name="/anomaly_map_generator/Gather_2" type="Gather" version="opset8">
1657
+ <data batch_dims="0" />
1658
+ <input>
1659
+ <port id="0" precision="I64">
1660
+ <dim>4</dim>
1661
+ </port>
1662
+ <port id="1" precision="I64">
1663
+ <dim>1</dim>
1664
+ </port>
1665
+ <port id="2" precision="I64" />
1666
+ </input>
1667
+ <output>
1668
+ <port id="3" precision="I64" names="/anomaly_map_generator/Gather_2_output_0,/anomaly_map_generator/Unsqueeze_4_output_0">
1669
+ <dim>1</dim>
1670
+ </port>
1671
+ </output>
1672
+ </layer>
1673
+ <layer id="92" name="Constant_2352" type="Const" version="opset1">
1674
+ <data element_type="i64" shape="1" offset="4009176" size="8" />
1675
+ <rt_info>
1676
+ <attribute name="precise" version="0" />
1677
+ </rt_info>
1678
+ <output>
1679
+ <port id="0" precision="I64">
1680
+ <dim>1</dim>
1681
+ </port>
1682
+ </output>
1683
+ </layer>
1684
+ <layer id="93" name="Constant_381" type="Const" version="opset1">
1685
+ <data element_type="i64" shape="" offset="993664" size="8" />
1686
+ <rt_info>
1687
+ <attribute name="precise" version="0" />
1688
+ </rt_info>
1689
+ <output>
1690
+ <port id="0" precision="I64" />
1691
+ </output>
1692
+ </layer>
1693
+ <layer id="94" name="/anomaly_map_generator/Gather_3" type="Gather" version="opset8">
1694
+ <data batch_dims="0" />
1695
+ <input>
1696
+ <port id="0" precision="I64">
1697
+ <dim>4</dim>
1698
+ </port>
1699
+ <port id="1" precision="I64">
1700
+ <dim>1</dim>
1701
+ </port>
1702
+ <port id="2" precision="I64" />
1703
+ </input>
1704
+ <output>
1705
+ <port id="3" precision="I64" names="/anomaly_map_generator/Gather_3_output_0,/anomaly_map_generator/Unsqueeze_5_output_0">
1706
+ <dim>1</dim>
1707
+ </port>
1708
+ </output>
1709
+ </layer>
1710
+ <layer id="95" name="/anomaly_map_generator/Mul" type="Multiply" version="opset1">
1711
+ <data auto_broadcast="numpy" />
1712
+ <input>
1713
+ <port id="0" precision="I64">
1714
+ <dim>1</dim>
1715
+ </port>
1716
+ <port id="1" precision="I64">
1717
+ <dim>1</dim>
1718
+ </port>
1719
+ </input>
1720
+ <output>
1721
+ <port id="2" precision="I64" names="/anomaly_map_generator/Mul_output_0,/anomaly_map_generator/Unsqueeze_2_output_0">
1722
+ <dim>1</dim>
1723
+ </port>
1724
+ </output>
1725
+ </layer>
1726
+ <layer id="96" name="/anomaly_map_generator/Concat" type="Concat" version="opset1">
1727
+ <data axis="0" />
1728
+ <input>
1729
+ <port id="0" precision="I64">
1730
+ <dim>2</dim>
1731
+ </port>
1732
+ <port id="1" precision="I64">
1733
+ <dim>1</dim>
1734
+ </port>
1735
+ </input>
1736
+ <output>
1737
+ <port id="2" precision="I64">
1738
+ <dim>3</dim>
1739
+ </port>
1740
+ </output>
1741
+ </layer>
1742
+ <layer id="97" name="/anomaly_map_generator/Reshape" type="Reshape" version="opset1">
1743
+ <data special_zero="true" />
1744
+ <input>
1745
+ <port id="0" precision="FP32">
1746
+ <dim>-1</dim>
1747
+ <dim>100</dim>
1748
+ <dim>64</dim>
1749
+ <dim>64</dim>
1750
+ </port>
1751
+ <port id="1" precision="I64">
1752
+ <dim>3</dim>
1753
+ </port>
1754
+ </input>
1755
+ <output>
1756
+ <port id="2" precision="FP32" names="/anomaly_map_generator/Reshape_output_0">
1757
+ <dim>-1</dim>
1758
+ <dim>100</dim>
1759
+ <dim>4096</dim>
1760
+ </port>
1761
+ </output>
1762
+ </layer>
1763
+ <layer id="98" name="Constant_2768_compressed" type="Const" version="opset1">
1764
+ <data element_type="f16" shape="1, 100, 4096" offset="4009184" size="819200" />
1765
+ <output>
1766
+ <port id="0" precision="FP16">
1767
+ <dim>1</dim>
1768
+ <dim>100</dim>
1769
+ <dim>4096</dim>
1770
+ </port>
1771
+ </output>
1772
+ </layer>
1773
+ <layer id="99" name="Constant_2768" type="Convert" version="opset1">
1774
+ <data destination_type="f32" />
1775
+ <rt_info>
1776
+ <attribute name="decompression" version="0" />
1777
+ </rt_info>
1778
+ <input>
1779
+ <port id="0" precision="FP16">
1780
+ <dim>1</dim>
1781
+ <dim>100</dim>
1782
+ <dim>4096</dim>
1783
+ </port>
1784
+ </input>
1785
+ <output>
1786
+ <port id="1" precision="FP32">
1787
+ <dim>1</dim>
1788
+ <dim>100</dim>
1789
+ <dim>4096</dim>
1790
+ </port>
1791
+ </output>
1792
+ </layer>
1793
+ <layer id="100" name="/anomaly_map_generator/Sub" type="Add" version="opset1">
1794
+ <data auto_broadcast="numpy" />
1795
+ <input>
1796
+ <port id="0" precision="FP32">
1797
+ <dim>-1</dim>
1798
+ <dim>100</dim>
1799
+ <dim>4096</dim>
1800
+ </port>
1801
+ <port id="1" precision="FP32">
1802
+ <dim>1</dim>
1803
+ <dim>100</dim>
1804
+ <dim>4096</dim>
1805
+ </port>
1806
+ </input>
1807
+ <output>
1808
+ <port id="2" precision="FP32" names="/anomaly_map_generator/Sub_output_0">
1809
+ <dim>-1</dim>
1810
+ <dim>100</dim>
1811
+ <dim>4096</dim>
1812
+ </port>
1813
+ </output>
1814
+ </layer>
1815
+ <layer id="101" name="Constant_397" type="Const" version="opset1">
1816
+ <data element_type="i64" shape="3" offset="4828384" size="24" />
1817
+ <output>
1818
+ <port id="0" precision="I64">
1819
+ <dim>3</dim>
1820
+ </port>
1821
+ </output>
1822
+ </layer>
1823
+ <layer id="102" name="/anomaly_map_generator/Transpose" type="Transpose" version="opset1">
1824
+ <input>
1825
+ <port id="0" precision="FP32">
1826
+ <dim>-1</dim>
1827
+ <dim>100</dim>
1828
+ <dim>4096</dim>
1829
+ </port>
1830
+ <port id="1" precision="I64">
1831
+ <dim>3</dim>
1832
+ </port>
1833
+ </input>
1834
+ <output>
1835
+ <port id="2" precision="FP32" names="/anomaly_map_generator/Transpose_output_0">
1836
+ <dim>4096</dim>
1837
+ <dim>-1</dim>
1838
+ <dim>100</dim>
1839
+ </port>
1840
+ </output>
1841
+ </layer>
1842
+ <layer id="103" name="onnx::MatMul_326_compressed" type="Const" version="opset1">
1843
+ <data element_type="f16" shape="4096, 100, 100" offset="4828408" size="81920000" />
1844
+ <output>
1845
+ <port id="0" precision="FP16" names="onnx::MatMul_326">
1846
+ <dim>4096</dim>
1847
+ <dim>100</dim>
1848
+ <dim>100</dim>
1849
+ </port>
1850
+ </output>
1851
+ </layer>
1852
+ <layer id="104" name="onnx::MatMul_326" type="Convert" version="opset1">
1853
+ <data destination_type="f32" />
1854
+ <rt_info>
1855
+ <attribute name="decompression" version="0" />
1856
+ </rt_info>
1857
+ <input>
1858
+ <port id="0" precision="FP16">
1859
+ <dim>4096</dim>
1860
+ <dim>100</dim>
1861
+ <dim>100</dim>
1862
+ </port>
1863
+ </input>
1864
+ <output>
1865
+ <port id="1" precision="FP32">
1866
+ <dim>4096</dim>
1867
+ <dim>100</dim>
1868
+ <dim>100</dim>
1869
+ </port>
1870
+ </output>
1871
+ </layer>
1872
+ <layer id="105" name="/anomaly_map_generator/MatMul" type="MatMul" version="opset1">
1873
+ <data transpose_a="false" transpose_b="false" />
1874
+ <input>
1875
+ <port id="0" precision="FP32">
1876
+ <dim>4096</dim>
1877
+ <dim>-1</dim>
1878
+ <dim>100</dim>
1879
+ </port>
1880
+ <port id="1" precision="FP32">
1881
+ <dim>4096</dim>
1882
+ <dim>100</dim>
1883
+ <dim>100</dim>
1884
+ </port>
1885
+ </input>
1886
+ <output>
1887
+ <port id="2" precision="FP32" names="/anomaly_map_generator/MatMul_output_0">
1888
+ <dim>4096</dim>
1889
+ <dim>-1</dim>
1890
+ <dim>100</dim>
1891
+ </port>
1892
+ </output>
1893
+ </layer>
1894
+ <layer id="106" name="/anomaly_map_generator/Mul_1" type="Multiply" version="opset1">
1895
+ <data auto_broadcast="numpy" />
1896
+ <input>
1897
+ <port id="0" precision="FP32">
1898
+ <dim>4096</dim>
1899
+ <dim>-1</dim>
1900
+ <dim>100</dim>
1901
+ </port>
1902
+ <port id="1" precision="FP32">
1903
+ <dim>4096</dim>
1904
+ <dim>-1</dim>
1905
+ <dim>100</dim>
1906
+ </port>
1907
+ </input>
1908
+ <output>
1909
+ <port id="2" precision="FP32" names="/anomaly_map_generator/Mul_1_output_0">
1910
+ <dim>4096</dim>
1911
+ <dim>-1</dim>
1912
+ <dim>100</dim>
1913
+ </port>
1914
+ </output>
1915
+ </layer>
1916
+ <layer id="107" name="Constant_401" type="Const" version="opset1">
1917
+ <data element_type="i64" shape="1" offset="993672" size="8" />
1918
+ <output>
1919
+ <port id="0" precision="I64">
1920
+ <dim>1</dim>
1921
+ </port>
1922
+ </output>
1923
+ </layer>
1924
+ <layer id="108" name="/anomaly_map_generator/ReduceSum" type="ReduceSum" version="opset1">
1925
+ <data keep_dims="false" />
1926
+ <input>
1927
+ <port id="0" precision="FP32">
1928
+ <dim>4096</dim>
1929
+ <dim>-1</dim>
1930
+ <dim>100</dim>
1931
+ </port>
1932
+ <port id="1" precision="I64">
1933
+ <dim>1</dim>
1934
+ </port>
1935
+ </input>
1936
+ <output>
1937
+ <port id="2" precision="FP32" names="/anomaly_map_generator/ReduceSum_output_0">
1938
+ <dim>4096</dim>
1939
+ <dim>-1</dim>
1940
+ </port>
1941
+ </output>
1942
+ </layer>
1943
+ <layer id="109" name="Constant_403" type="Const" version="opset1">
1944
+ <data element_type="i64" shape="2" offset="86748408" size="16" />
1945
+ <output>
1946
+ <port id="0" precision="I64">
1947
+ <dim>2</dim>
1948
+ </port>
1949
+ </output>
1950
+ </layer>
1951
+ <layer id="110" name="/anomaly_map_generator/Transpose_1" type="Transpose" version="opset1">
1952
+ <input>
1953
+ <port id="0" precision="FP32">
1954
+ <dim>4096</dim>
1955
+ <dim>-1</dim>
1956
+ </port>
1957
+ <port id="1" precision="I64">
1958
+ <dim>2</dim>
1959
+ </port>
1960
+ </input>
1961
+ <output>
1962
+ <port id="2" precision="FP32" names="/anomaly_map_generator/Transpose_1_output_0">
1963
+ <dim>-1</dim>
1964
+ <dim>4096</dim>
1965
+ </port>
1966
+ </output>
1967
+ </layer>
1968
+ <layer id="111" name="Constant_2343" type="Const" version="opset1">
1969
+ <data element_type="i64" shape="1" offset="993664" size="8" />
1970
+ <rt_info>
1971
+ <attribute name="precise" version="0" />
1972
+ </rt_info>
1973
+ <output>
1974
+ <port id="0" precision="I64">
1975
+ <dim>1</dim>
1976
+ </port>
1977
+ </output>
1978
+ </layer>
1979
+ <layer id="112" name="Constant_369" type="Const" version="opset1">
1980
+ <data element_type="i64" shape="" offset="993664" size="8" />
1981
+ <rt_info>
1982
+ <attribute name="precise" version="0" />
1983
+ </rt_info>
1984
+ <output>
1985
+ <port id="0" precision="I64" />
1986
+ </output>
1987
+ </layer>
1988
+ <layer id="113" name="/anomaly_map_generator/Gather" type="Gather" version="opset8">
1989
+ <data batch_dims="0" />
1990
+ <input>
1991
+ <port id="0" precision="I64">
1992
+ <dim>4</dim>
1993
+ </port>
1994
+ <port id="1" precision="I64">
1995
+ <dim>1</dim>
1996
+ </port>
1997
+ <port id="2" precision="I64" />
1998
+ </input>
1999
+ <output>
2000
+ <port id="3" precision="I64" names="/anomaly_map_generator/Gather_output_0,/anomaly_map_generator/Unsqueeze_3_output_0,/anomaly_map_generator/Unsqueeze_output_0">
2001
+ <dim>1</dim>
2002
+ </port>
2003
+ </output>
2004
+ </layer>
2005
+ <layer id="114" name="/anomaly_map_generator/Constant_4" type="Const" version="opset1">
2006
+ <data element_type="i64" shape="1" offset="993680" size="8" />
2007
+ <rt_info>
2008
+ <attribute name="precise" version="0" />
2009
+ </rt_info>
2010
+ <output>
2011
+ <port id="0" precision="I64" names="/anomaly_map_generator/Constant_4_output_0">
2012
+ <dim>1</dim>
2013
+ </port>
2014
+ </output>
2015
+ </layer>
2016
+ <layer id="115" name="Constant_2379" type="Const" version="opset1">
2017
+ <data element_type="i64" shape="2" offset="993688" size="16" />
2018
+ <rt_info>
2019
+ <attribute name="precise" version="0" />
2020
+ </rt_info>
2021
+ <output>
2022
+ <port id="0" precision="I64">
2023
+ <dim>2</dim>
2024
+ </port>
2025
+ </output>
2026
+ </layer>
2027
+ <layer id="116" name="Constant_2380" type="Const" version="opset1">
2028
+ <data element_type="i64" shape="" offset="993664" size="8" />
2029
+ <rt_info>
2030
+ <attribute name="precise" version="0" />
2031
+ </rt_info>
2032
+ <output>
2033
+ <port id="0" precision="I64" />
2034
+ </output>
2035
+ </layer>
2036
+ <layer id="117" name="Gather_2381" type="Gather" version="opset8">
2037
+ <data batch_dims="0" />
2038
+ <input>
2039
+ <port id="0" precision="I64">
2040
+ <dim>4</dim>
2041
+ </port>
2042
+ <port id="1" precision="I64">
2043
+ <dim>2</dim>
2044
+ </port>
2045
+ <port id="2" precision="I64" />
2046
+ </input>
2047
+ <output>
2048
+ <port id="3" precision="I64">
2049
+ <dim>2</dim>
2050
+ </port>
2051
+ </output>
2052
+ </layer>
2053
+ <layer id="118" name="/anomaly_map_generator/Concat_1" type="Concat" version="opset1">
2054
+ <data axis="0" />
2055
+ <input>
2056
+ <port id="0" precision="I64">
2057
+ <dim>1</dim>
2058
+ </port>
2059
+ <port id="1" precision="I64">
2060
+ <dim>1</dim>
2061
+ </port>
2062
+ <port id="2" precision="I64">
2063
+ <dim>2</dim>
2064
+ </port>
2065
+ </input>
2066
+ <output>
2067
+ <port id="3" precision="I64" names="/anomaly_map_generator/Concat_1_output_0">
2068
+ <dim>4</dim>
2069
+ </port>
2070
+ </output>
2071
+ </layer>
2072
+ <layer id="119" name="/anomaly_map_generator/Reshape_1" type="Reshape" version="opset1">
2073
+ <data special_zero="true" />
2074
+ <input>
2075
+ <port id="0" precision="FP32">
2076
+ <dim>-1</dim>
2077
+ <dim>4096</dim>
2078
+ </port>
2079
+ <port id="1" precision="I64">
2080
+ <dim>4</dim>
2081
+ </port>
2082
+ </input>
2083
+ <output>
2084
+ <port id="2" precision="FP32" names="/anomaly_map_generator/Reshape_1_output_0">
2085
+ <dim>-1</dim>
2086
+ <dim>1</dim>
2087
+ <dim>64</dim>
2088
+ <dim>64</dim>
2089
+ </port>
2090
+ </output>
2091
+ </layer>
2092
+ <layer id="120" name="/anomaly_map_generator/Clip" type="Clamp" version="opset1">
2093
+ <data min="0" max="3.4028234663852886e+38" />
2094
+ <input>
2095
+ <port id="0" precision="FP32">
2096
+ <dim>-1</dim>
2097
+ <dim>1</dim>
2098
+ <dim>64</dim>
2099
+ <dim>64</dim>
2100
+ </port>
2101
+ </input>
2102
+ <output>
2103
+ <port id="1" precision="FP32" names="/anomaly_map_generator/Clip_output_0">
2104
+ <dim>-1</dim>
2105
+ <dim>1</dim>
2106
+ <dim>64</dim>
2107
+ <dim>64</dim>
2108
+ </port>
2109
+ </output>
2110
+ </layer>
2111
+ <layer id="121" name="/anomaly_map_generator/Sqrt" type="Sqrt" version="opset1">
2112
+ <input>
2113
+ <port id="0" precision="FP32">
2114
+ <dim>-1</dim>
2115
+ <dim>1</dim>
2116
+ <dim>64</dim>
2117
+ <dim>64</dim>
2118
+ </port>
2119
+ </input>
2120
+ <output>
2121
+ <port id="1" precision="FP32" names="/anomaly_map_generator/Sqrt_output_0">
2122
+ <dim>-1</dim>
2123
+ <dim>1</dim>
2124
+ <dim>64</dim>
2125
+ <dim>64</dim>
2126
+ </port>
2127
+ </output>
2128
+ </layer>
2129
+ <layer id="122" name="/anomaly_map_generator/Shape_4" type="ShapeOf" version="opset3">
2130
+ <data output_type="i64" />
2131
+ <input>
2132
+ <port id="0" precision="FP32">
2133
+ <dim>-1</dim>
2134
+ <dim>1</dim>
2135
+ <dim>64</dim>
2136
+ <dim>64</dim>
2137
+ </port>
2138
+ </input>
2139
+ <output>
2140
+ <port id="1" precision="I64" names="/anomaly_map_generator/Shape_4_output_0">
2141
+ <dim>4</dim>
2142
+ </port>
2143
+ </output>
2144
+ </layer>
2145
+ <layer id="123" name="/anomaly_map_generator/Constant_7" type="Const" version="opset1">
2146
+ <data element_type="i64" shape="1" offset="993664" size="8" />
2147
+ <rt_info>
2148
+ <attribute name="precise" version="0" />
2149
+ </rt_info>
2150
+ <output>
2151
+ <port id="0" precision="I64" names="/anomaly_map_generator/Constant_7_output_0">
2152
+ <dim>1</dim>
2153
+ </port>
2154
+ </output>
2155
+ </layer>
2156
+ <layer id="124" name="/anomaly_map_generator/Constant_8" type="Const" version="opset1">
2157
+ <data element_type="i64" shape="1" offset="993672" size="8" />
2158
+ <rt_info>
2159
+ <attribute name="precise" version="0" />
2160
+ </rt_info>
2161
+ <output>
2162
+ <port id="0" precision="I64" names="/anomaly_map_generator/Constant_8_output_0">
2163
+ <dim>1</dim>
2164
+ </port>
2165
+ </output>
2166
+ </layer>
2167
+ <layer id="125" name="Broadcast_429" type="Const" version="opset1">
2168
+ <data element_type="i64" shape="1" offset="993680" size="8" />
2169
+ <rt_info>
2170
+ <attribute name="precise" version="0" />
2171
+ </rt_info>
2172
+ <output>
2173
+ <port id="0" precision="I64">
2174
+ <dim>1</dim>
2175
+ </port>
2176
+ </output>
2177
+ </layer>
2178
+ <layer id="126" name="/anomaly_map_generator/Slice" type="StridedSlice" version="opset1">
2179
+ <data begin_mask="0" end_mask="0" new_axis_mask="" shrink_axis_mask="" ellipsis_mask="" />
2180
+ <input>
2181
+ <port id="0" precision="I64">
2182
+ <dim>4</dim>
2183
+ </port>
2184
+ <port id="1" precision="I64">
2185
+ <dim>1</dim>
2186
+ </port>
2187
+ <port id="2" precision="I64">
2188
+ <dim>1</dim>
2189
+ </port>
2190
+ <port id="3" precision="I64">
2191
+ <dim>1</dim>
2192
+ </port>
2193
+ </input>
2194
+ <output>
2195
+ <port id="4" precision="I64" names="/anomaly_map_generator/Slice_output_0">
2196
+ <dim>2</dim>
2197
+ </port>
2198
+ </output>
2199
+ </layer>
2200
+ <layer id="127" name="/anomaly_map_generator/Constant_9" type="Const" version="opset1">
2201
+ <data element_type="i64" shape="2" offset="86748424" size="16" />
2202
+ <rt_info>
2203
+ <attribute name="precise" version="0" />
2204
+ </rt_info>
2205
+ <output>
2206
+ <port id="0" precision="I64" names="/anomaly_map_generator/Constant_9_output_0">
2207
+ <dim>2</dim>
2208
+ </port>
2209
+ </output>
2210
+ </layer>
2211
+ <layer id="128" name="/anomaly_map_generator/Concat_2" type="Concat" version="opset1">
2212
+ <data axis="0" />
2213
+ <input>
2214
+ <port id="0" precision="I64">
2215
+ <dim>2</dim>
2216
+ </port>
2217
+ <port id="1" precision="I64">
2218
+ <dim>2</dim>
2219
+ </port>
2220
+ </input>
2221
+ <output>
2222
+ <port id="2" precision="I64" names="/anomaly_map_generator/Concat_2_output_0">
2223
+ <dim>4</dim>
2224
+ </port>
2225
+ </output>
2226
+ </layer>
2227
+ <layer id="129" name="/anomaly_map_generator/Resize" type="Interpolate" version="opset11">
2228
+ <data mode="linear_onnx" shape_calculation_mode="sizes" coordinate_transformation_mode="half_pixel" nearest_mode="floor" antialias="false" pads_begin="0, 0, 0, 0" pads_end="0, 0, 0, 0" cube_coeff="-0.75" />
2229
+ <input>
2230
+ <port id="0" precision="FP32">
2231
+ <dim>-1</dim>
2232
+ <dim>1</dim>
2233
+ <dim>64</dim>
2234
+ <dim>64</dim>
2235
+ </port>
2236
+ <port id="1" precision="I64">
2237
+ <dim>4</dim>
2238
+ </port>
2239
+ </input>
2240
+ <output>
2241
+ <port id="2" precision="FP32" names="/anomaly_map_generator/Resize_output_0">
2242
+ <dim>-1</dim>
2243
+ <dim>1</dim>
2244
+ <dim>256</dim>
2245
+ <dim>256</dim>
2246
+ </port>
2247
+ </output>
2248
+ </layer>
2249
+ <layer id="130" name="Split_481.0" type="Const" version="opset1">
2250
+ <data element_type="i64" shape="4" offset="86748440" size="32" />
2251
+ <rt_info>
2252
+ <attribute name="precise" version="0" />
2253
+ </rt_info>
2254
+ <output>
2255
+ <port id="0" precision="I64">
2256
+ <dim>4</dim>
2257
+ </port>
2258
+ </output>
2259
+ </layer>
2260
+ <layer id="131" name="Split_481.1" type="Const" version="opset1">
2261
+ <data element_type="i64" shape="4" offset="86748440" size="32" />
2262
+ <rt_info>
2263
+ <attribute name="precise" version="0" />
2264
+ </rt_info>
2265
+ <output>
2266
+ <port id="0" precision="I64">
2267
+ <dim>4</dim>
2268
+ </port>
2269
+ </output>
2270
+ </layer>
2271
+ <layer id="132" name="Constant_479_compressed" type="Const" version="opset1">
2272
+ <data element_type="f16" shape="" offset="86748472" size="2" />
2273
+ <output>
2274
+ <port id="0" precision="FP16" />
2275
+ </output>
2276
+ </layer>
2277
+ <layer id="133" name="Constant_479" type="Convert" version="opset1">
2278
+ <data destination_type="f32" />
2279
+ <rt_info>
2280
+ <attribute name="decompression" version="0" />
2281
+ </rt_info>
2282
+ <input>
2283
+ <port id="0" precision="FP16" />
2284
+ </input>
2285
+ <output>
2286
+ <port id="1" precision="FP32" />
2287
+ </output>
2288
+ </layer>
2289
+ <layer id="134" name="/anomaly_map_generator/blur/Pad" type="Pad" version="opset12">
2290
+ <data pad_mode="reflect" />
2291
+ <input>
2292
+ <port id="0" precision="FP32">
2293
+ <dim>-1</dim>
2294
+ <dim>1</dim>
2295
+ <dim>256</dim>
2296
+ <dim>256</dim>
2297
+ </port>
2298
+ <port id="1" precision="I64">
2299
+ <dim>4</dim>
2300
+ </port>
2301
+ <port id="2" precision="I64">
2302
+ <dim>4</dim>
2303
+ </port>
2304
+ <port id="3" precision="FP32" />
2305
+ </input>
2306
+ <output>
2307
+ <port id="4" precision="FP32" names="/anomaly_map_generator/blur/Pad_output_0">
2308
+ <dim>-1</dim>
2309
+ <dim>1</dim>
2310
+ <dim>288</dim>
2311
+ <dim>288</dim>
2312
+ </port>
2313
+ </output>
2314
+ </layer>
2315
+ <layer id="135" name="anomaly_map_generator.blur.kernel_compressed" type="Const" version="opset1">
2316
+ <data element_type="f16" shape="1, 1, 33, 33" offset="86748474" size="2178" />
2317
+ <output>
2318
+ <port id="0" precision="FP16" names="anomaly_map_generator.blur.kernel">
2319
+ <dim>1</dim>
2320
+ <dim>1</dim>
2321
+ <dim>33</dim>
2322
+ <dim>33</dim>
2323
+ </port>
2324
+ </output>
2325
+ </layer>
2326
+ <layer id="136" name="anomaly_map_generator.blur.kernel" type="Convert" version="opset1">
2327
+ <data destination_type="f32" />
2328
+ <rt_info>
2329
+ <attribute name="decompression" version="0" />
2330
+ </rt_info>
2331
+ <input>
2332
+ <port id="0" precision="FP16">
2333
+ <dim>1</dim>
2334
+ <dim>1</dim>
2335
+ <dim>33</dim>
2336
+ <dim>33</dim>
2337
+ </port>
2338
+ </input>
2339
+ <output>
2340
+ <port id="1" precision="FP32">
2341
+ <dim>1</dim>
2342
+ <dim>1</dim>
2343
+ <dim>33</dim>
2344
+ <dim>33</dim>
2345
+ </port>
2346
+ </output>
2347
+ </layer>
2348
+ <layer id="137" name="/anomaly_map_generator/blur/Conv" type="Convolution" version="opset1">
2349
+ <data strides="1, 1" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" auto_pad="explicit" />
2350
+ <input>
2351
+ <port id="0" precision="FP32">
2352
+ <dim>-1</dim>
2353
+ <dim>1</dim>
2354
+ <dim>288</dim>
2355
+ <dim>288</dim>
2356
+ </port>
2357
+ <port id="1" precision="FP32">
2358
+ <dim>1</dim>
2359
+ <dim>1</dim>
2360
+ <dim>33</dim>
2361
+ <dim>33</dim>
2362
+ </port>
2363
+ </input>
2364
+ <output>
2365
+ <port id="2" precision="FP32" names="/anomaly_map_generator/blur/Conv_output_0">
2366
+ <dim>-1</dim>
2367
+ <dim>1</dim>
2368
+ <dim>256</dim>
2369
+ <dim>256</dim>
2370
+ </port>
2371
+ </output>
2372
+ </layer>
2373
+ <layer id="138" name="/anomaly_map_generator/blur/Shape" type="ShapeOf" version="opset3">
2374
+ <data output_type="i64" />
2375
+ <input>
2376
+ <port id="0" precision="FP32">
2377
+ <dim>-1</dim>
2378
+ <dim>1</dim>
2379
+ <dim>256</dim>
2380
+ <dim>256</dim>
2381
+ </port>
2382
+ </input>
2383
+ <output>
2384
+ <port id="1" precision="I64" names="/anomaly_map_generator/blur/Concat_1_output_0,/anomaly_map_generator/blur/Shape_1_output_0,/anomaly_map_generator/blur/Shape_2_output_0,/anomaly_map_generator/blur/Shape_3_output_0,/anomaly_map_generator/blur/Shape_output_0">
2385
+ <dim>4</dim>
2386
+ </port>
2387
+ </output>
2388
+ </layer>
2389
+ <layer id="139" name="output" type="Reshape" version="opset1">
2390
+ <data special_zero="true" />
2391
+ <input>
2392
+ <port id="0" precision="FP32">
2393
+ <dim>-1</dim>
2394
+ <dim>1</dim>
2395
+ <dim>256</dim>
2396
+ <dim>256</dim>
2397
+ </port>
2398
+ <port id="1" precision="I64">
2399
+ <dim>4</dim>
2400
+ </port>
2401
+ </input>
2402
+ <output>
2403
+ <port id="2" precision="FP32" names="output">
2404
+ <dim>-1</dim>
2405
+ <dim>1</dim>
2406
+ <dim>256</dim>
2407
+ <dim>256</dim>
2408
+ </port>
2409
+ </output>
2410
+ </layer>
2411
+ <layer id="140" name="output/sink_port_0" type="Result" version="opset1">
2412
+ <input>
2413
+ <port id="0" precision="FP32">
2414
+ <dim>-1</dim>
2415
+ <dim>1</dim>
2416
+ <dim>256</dim>
2417
+ <dim>256</dim>
2418
+ </port>
2419
+ </input>
2420
+ </layer>
2421
+ </layers>
2422
+ <edges>
2423
+ <edge from-layer="0" from-port="0" to-layer="3" to-port="0" />
2424
+ <edge from-layer="1" from-port="0" to-layer="2" to-port="0" />
2425
+ <edge from-layer="2" from-port="1" to-layer="3" to-port="1" />
2426
+ <edge from-layer="3" from-port="2" to-layer="4" to-port="0" />
2427
+ <edge from-layer="4" from-port="1" to-layer="5" to-port="0" />
2428
+ <edge from-layer="5" from-port="1" to-layer="8" to-port="0" />
2429
+ <edge from-layer="5" from-port="1" to-layer="13" to-port="1" />
2430
+ <edge from-layer="6" from-port="0" to-layer="7" to-port="0" />
2431
+ <edge from-layer="7" from-port="1" to-layer="8" to-port="1" />
2432
+ <edge from-layer="8" from-port="2" to-layer="9" to-port="0" />
2433
+ <edge from-layer="9" from-port="1" to-layer="12" to-port="0" />
2434
+ <edge from-layer="10" from-port="0" to-layer="11" to-port="0" />
2435
+ <edge from-layer="11" from-port="1" to-layer="12" to-port="1" />
2436
+ <edge from-layer="11" from-port="1" to-layer="19" to-port="1" />
2437
+ <edge from-layer="12" from-port="2" to-layer="13" to-port="0" />
2438
+ <edge from-layer="13" from-port="2" to-layer="14" to-port="0" />
2439
+ <edge from-layer="14" from-port="1" to-layer="17" to-port="0" />
2440
+ <edge from-layer="14" from-port="1" to-layer="20" to-port="1" />
2441
+ <edge from-layer="15" from-port="0" to-layer="16" to-port="0" />
2442
+ <edge from-layer="16" from-port="1" to-layer="17" to-port="1" />
2443
+ <edge from-layer="17" from-port="2" to-layer="18" to-port="0" />
2444
+ <edge from-layer="18" from-port="1" to-layer="19" to-port="0" />
2445
+ <edge from-layer="19" from-port="2" to-layer="20" to-port="0" />
2446
+ <edge from-layer="20" from-port="2" to-layer="21" to-port="0" />
2447
+ <edge from-layer="21" from-port="1" to-layer="24" to-port="0" />
2448
+ <edge from-layer="21" from-port="1" to-layer="31" to-port="0" />
2449
+ <edge from-layer="21" from-port="1" to-layer="46" to-port="0" />
2450
+ <edge from-layer="21" from-port="1" to-layer="52" to-port="0" />
2451
+ <edge from-layer="22" from-port="0" to-layer="23" to-port="0" />
2452
+ <edge from-layer="23" from-port="1" to-layer="24" to-port="1" />
2453
+ <edge from-layer="24" from-port="2" to-layer="25" to-port="0" />
2454
+ <edge from-layer="25" from-port="1" to-layer="28" to-port="0" />
2455
+ <edge from-layer="26" from-port="0" to-layer="27" to-port="0" />
2456
+ <edge from-layer="27" from-port="1" to-layer="28" to-port="1" />
2457
+ <edge from-layer="27" from-port="1" to-layer="38" to-port="1" />
2458
+ <edge from-layer="28" from-port="2" to-layer="32" to-port="0" />
2459
+ <edge from-layer="29" from-port="0" to-layer="30" to-port="0" />
2460
+ <edge from-layer="30" from-port="1" to-layer="31" to-port="1" />
2461
+ <edge from-layer="31" from-port="2" to-layer="32" to-port="1" />
2462
+ <edge from-layer="32" from-port="2" to-layer="33" to-port="0" />
2463
+ <edge from-layer="33" from-port="1" to-layer="36" to-port="0" />
2464
+ <edge from-layer="33" from-port="1" to-layer="39" to-port="1" />
2465
+ <edge from-layer="34" from-port="0" to-layer="35" to-port="0" />
2466
+ <edge from-layer="35" from-port="1" to-layer="36" to-port="1" />
2467
+ <edge from-layer="36" from-port="2" to-layer="37" to-port="0" />
2468
+ <edge from-layer="37" from-port="1" to-layer="38" to-port="0" />
2469
+ <edge from-layer="38" from-port="2" to-layer="39" to-port="0" />
2470
+ <edge from-layer="39" from-port="2" to-layer="40" to-port="0" />
2471
+ <edge from-layer="40" from-port="1" to-layer="41" to-port="0" />
2472
+ <edge from-layer="40" from-port="1" to-layer="51" to-port="0" />
2473
+ <edge from-layer="40" from-port="1" to-layer="55" to-port="0" />
2474
+ <edge from-layer="40" from-port="1" to-layer="62" to-port="0" />
2475
+ <edge from-layer="41" from-port="1" to-layer="45" to-port="0" />
2476
+ <edge from-layer="42" from-port="0" to-layer="45" to-port="1" />
2477
+ <edge from-layer="43" from-port="0" to-layer="45" to-port="2" />
2478
+ <edge from-layer="44" from-port="0" to-layer="45" to-port="3" />
2479
+ <edge from-layer="45" from-port="4" to-layer="50" to-port="0" />
2480
+ <edge from-layer="46" from-port="1" to-layer="49" to-port="0" />
2481
+ <edge from-layer="47" from-port="0" to-layer="49" to-port="1" />
2482
+ <edge from-layer="48" from-port="0" to-layer="49" to-port="2" />
2483
+ <edge from-layer="49" from-port="3" to-layer="50" to-port="1" />
2484
+ <edge from-layer="50" from-port="2" to-layer="51" to-port="1" />
2485
+ <edge from-layer="51" from-port="2" to-layer="52" to-port="1" />
2486
+ <edge from-layer="52" from-port="2" to-layer="77" to-port="0" />
2487
+ <edge from-layer="52" from-port="2" to-layer="83" to-port="0" />
2488
+ <edge from-layer="53" from-port="0" to-layer="54" to-port="0" />
2489
+ <edge from-layer="54" from-port="1" to-layer="55" to-port="1" />
2490
+ <edge from-layer="55" from-port="2" to-layer="56" to-port="0" />
2491
+ <edge from-layer="56" from-port="1" to-layer="59" to-port="0" />
2492
+ <edge from-layer="57" from-port="0" to-layer="58" to-port="0" />
2493
+ <edge from-layer="58" from-port="1" to-layer="59" to-port="1" />
2494
+ <edge from-layer="58" from-port="1" to-layer="69" to-port="1" />
2495
+ <edge from-layer="59" from-port="2" to-layer="63" to-port="0" />
2496
+ <edge from-layer="60" from-port="0" to-layer="61" to-port="0" />
2497
+ <edge from-layer="61" from-port="1" to-layer="62" to-port="1" />
2498
+ <edge from-layer="62" from-port="2" to-layer="63" to-port="1" />
2499
+ <edge from-layer="63" from-port="2" to-layer="64" to-port="0" />
2500
+ <edge from-layer="64" from-port="1" to-layer="70" to-port="1" />
2501
+ <edge from-layer="64" from-port="1" to-layer="67" to-port="0" />
2502
+ <edge from-layer="65" from-port="0" to-layer="66" to-port="0" />
2503
+ <edge from-layer="66" from-port="1" to-layer="67" to-port="1" />
2504
+ <edge from-layer="67" from-port="2" to-layer="68" to-port="0" />
2505
+ <edge from-layer="68" from-port="1" to-layer="69" to-port="0" />
2506
+ <edge from-layer="69" from-port="2" to-layer="70" to-port="0" />
2507
+ <edge from-layer="70" from-port="2" to-layer="71" to-port="0" />
2508
+ <edge from-layer="71" from-port="1" to-layer="72" to-port="0" />
2509
+ <edge from-layer="71" from-port="1" to-layer="82" to-port="0" />
2510
+ <edge from-layer="72" from-port="1" to-layer="76" to-port="0" />
2511
+ <edge from-layer="73" from-port="0" to-layer="76" to-port="1" />
2512
+ <edge from-layer="74" from-port="0" to-layer="76" to-port="2" />
2513
+ <edge from-layer="75" from-port="0" to-layer="76" to-port="3" />
2514
+ <edge from-layer="76" from-port="4" to-layer="81" to-port="0" />
2515
+ <edge from-layer="77" from-port="1" to-layer="80" to-port="0" />
2516
+ <edge from-layer="78" from-port="0" to-layer="80" to-port="1" />
2517
+ <edge from-layer="79" from-port="0" to-layer="80" to-port="2" />
2518
+ <edge from-layer="80" from-port="3" to-layer="81" to-port="1" />
2519
+ <edge from-layer="81" from-port="2" to-layer="82" to-port="1" />
2520
+ <edge from-layer="82" from-port="2" to-layer="83" to-port="1" />
2521
+ <edge from-layer="83" from-port="2" to-layer="86" to-port="0" />
2522
+ <edge from-layer="84" from-port="0" to-layer="86" to-port="1" />
2523
+ <edge from-layer="85" from-port="0" to-layer="86" to-port="2" />
2524
+ <edge from-layer="86" from-port="3" to-layer="88" to-port="0" />
2525
+ <edge from-layer="86" from-port="3" to-layer="97" to-port="0" />
2526
+ <edge from-layer="87" from-port="0" to-layer="96" to-port="0" />
2527
+ <edge from-layer="88" from-port="1" to-layer="91" to-port="0" />
2528
+ <edge from-layer="88" from-port="1" to-layer="94" to-port="0" />
2529
+ <edge from-layer="88" from-port="1" to-layer="117" to-port="0" />
2530
+ <edge from-layer="88" from-port="1" to-layer="113" to-port="0" />
2531
+ <edge from-layer="89" from-port="0" to-layer="91" to-port="1" />
2532
+ <edge from-layer="90" from-port="0" to-layer="91" to-port="2" />
2533
+ <edge from-layer="91" from-port="3" to-layer="95" to-port="0" />
2534
+ <edge from-layer="92" from-port="0" to-layer="94" to-port="1" />
2535
+ <edge from-layer="93" from-port="0" to-layer="94" to-port="2" />
2536
+ <edge from-layer="94" from-port="3" to-layer="95" to-port="1" />
2537
+ <edge from-layer="95" from-port="2" to-layer="96" to-port="1" />
2538
+ <edge from-layer="96" from-port="2" to-layer="97" to-port="1" />
2539
+ <edge from-layer="97" from-port="2" to-layer="100" to-port="0" />
2540
+ <edge from-layer="98" from-port="0" to-layer="99" to-port="0" />
2541
+ <edge from-layer="99" from-port="1" to-layer="100" to-port="1" />
2542
+ <edge from-layer="100" from-port="2" to-layer="102" to-port="0" />
2543
+ <edge from-layer="101" from-port="0" to-layer="102" to-port="1" />
2544
+ <edge from-layer="102" from-port="2" to-layer="106" to-port="1" />
2545
+ <edge from-layer="102" from-port="2" to-layer="105" to-port="0" />
2546
+ <edge from-layer="103" from-port="0" to-layer="104" to-port="0" />
2547
+ <edge from-layer="104" from-port="1" to-layer="105" to-port="1" />
2548
+ <edge from-layer="105" from-port="2" to-layer="106" to-port="0" />
2549
+ <edge from-layer="106" from-port="2" to-layer="108" to-port="0" />
2550
+ <edge from-layer="107" from-port="0" to-layer="108" to-port="1" />
2551
+ <edge from-layer="108" from-port="2" to-layer="110" to-port="0" />
2552
+ <edge from-layer="109" from-port="0" to-layer="110" to-port="1" />
2553
+ <edge from-layer="110" from-port="2" to-layer="119" to-port="0" />
2554
+ <edge from-layer="111" from-port="0" to-layer="113" to-port="1" />
2555
+ <edge from-layer="112" from-port="0" to-layer="113" to-port="2" />
2556
+ <edge from-layer="113" from-port="3" to-layer="118" to-port="0" />
2557
+ <edge from-layer="114" from-port="0" to-layer="118" to-port="1" />
2558
+ <edge from-layer="115" from-port="0" to-layer="117" to-port="1" />
2559
+ <edge from-layer="116" from-port="0" to-layer="117" to-port="2" />
2560
+ <edge from-layer="117" from-port="3" to-layer="118" to-port="2" />
2561
+ <edge from-layer="118" from-port="3" to-layer="119" to-port="1" />
2562
+ <edge from-layer="119" from-port="2" to-layer="120" to-port="0" />
2563
+ <edge from-layer="120" from-port="1" to-layer="121" to-port="0" />
2564
+ <edge from-layer="121" from-port="1" to-layer="122" to-port="0" />
2565
+ <edge from-layer="121" from-port="1" to-layer="129" to-port="0" />
2566
+ <edge from-layer="122" from-port="1" to-layer="126" to-port="0" />
2567
+ <edge from-layer="123" from-port="0" to-layer="126" to-port="1" />
2568
+ <edge from-layer="124" from-port="0" to-layer="126" to-port="2" />
2569
+ <edge from-layer="125" from-port="0" to-layer="126" to-port="3" />
2570
+ <edge from-layer="126" from-port="4" to-layer="128" to-port="0" />
2571
+ <edge from-layer="127" from-port="0" to-layer="128" to-port="1" />
2572
+ <edge from-layer="128" from-port="2" to-layer="129" to-port="1" />
2573
+ <edge from-layer="129" from-port="2" to-layer="134" to-port="0" />
2574
+ <edge from-layer="129" from-port="2" to-layer="138" to-port="0" />
2575
+ <edge from-layer="130" from-port="0" to-layer="134" to-port="1" />
2576
+ <edge from-layer="131" from-port="0" to-layer="134" to-port="2" />
2577
+ <edge from-layer="132" from-port="0" to-layer="133" to-port="0" />
2578
+ <edge from-layer="133" from-port="1" to-layer="134" to-port="3" />
2579
+ <edge from-layer="134" from-port="4" to-layer="137" to-port="0" />
2580
+ <edge from-layer="135" from-port="0" to-layer="136" to-port="0" />
2581
+ <edge from-layer="136" from-port="1" to-layer="137" to-port="1" />
2582
+ <edge from-layer="137" from-port="2" to-layer="139" to-port="0" />
2583
+ <edge from-layer="138" from-port="1" to-layer="139" to-port="1" />
2584
+ <edge from-layer="139" from-port="2" to-layer="140" to-port="0" />
2585
+ </edges>
2586
+ <rt_info>
2587
+ <Runtime_version value="2023.3.0-13775-ceeafaf64f3-releases/2023/3" />
2588
+ <conversion_parameters>
2589
+ <is_python_object value="False" />
2590
+ </conversion_parameters>
2591
+ <model_info>
2592
+ <image_shape value="256 256" />
2593
+ <image_threshold value="1.86658" />
2594
+ <labels value="Normal Anomaly" />
2595
+ <mean_values value="123.675 116.28 103.53" />
2596
+ <model_type value="AnomalyDetection" />
2597
+ <normalization_scale value="33.0247" />
2598
+ <orig_height value="256" />
2599
+ <orig_width value="341" />
2600
+ <pixel_threshold value="1.59375" />
2601
+ <reverse_input_channels value="YES" />
2602
+ <scale_values value="58.395 57.120000000000005 57.375" />
2603
+ <task value="segmentation" />
2604
+ </model_info>
2605
+ </rt_info>
2606
+ </net>