Datasets:
Update README.md
Browse files
README.md
CHANGED
@@ -33,14 +33,33 @@ size_categories:
|
|
33 |
This dataset is a subset of Kaggle's Google Landmark Recognition 2021 competition with only the categories with more than 500 images.
|
34 |
https://www.kaggle.com/competitions/landmark-recognition-2021/data
|
35 |
|
36 |
-
The dataset consists of a total of 45579 224x224 color images in 51
|
37 |
-
|
38 |
-
The dataset was randomly split with 80% of the images for the train set and 20% for the test set.
|
39 |
|
40 |
### Languages
|
41 |
|
42 |
English
|
43 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
44 |
### Source Data
|
45 |
|
46 |
The full dataset is from Kaggle Landmark Recognition 2021
|
@@ -54,32 +73,3 @@ https://www.kaggle.com/competitions/landmark-recognition-2021/data
|
|
54 |
"Google Landmarks Dataset v2 - A Large-Scale Benchmark for Instance-Level Recognition and Retrieval", T. Weyand, A. Araujo, B. Cao and J. Sim, Proc. CVPR'20
|
55 |
|
56 |
"Towards A Fairer Landmark Recognition Dataset", Z. Kim, A. Araujo, B. Cao, C. Askew, J. Sim, M. Green, N. Yilla and T. Weyand, arxiv:2108.08874
|
57 |
-
|
58 |
-
### Dataset info
|
59 |
-
|
60 |
-
dataset_info:
|
61 |
-
features:
|
62 |
-
- name: id
|
63 |
-
dtype: string
|
64 |
-
- name: landmark_id
|
65 |
-
dtype: int64
|
66 |
-
- name: category
|
67 |
-
dtype: string
|
68 |
-
- name: image
|
69 |
-
dtype: image
|
70 |
-
- name: label
|
71 |
-
dtype: int64
|
72 |
-
splits:
|
73 |
-
- name: train
|
74 |
-
num_bytes: 2428986323.125
|
75 |
-
num_examples: 36463
|
76 |
-
- name: test
|
77 |
-
num_bytes: 606874794.5
|
78 |
-
num_examples: 9116
|
79 |
-
download_size: 3034360629
|
80 |
-
dataset_size: 3035861117.625
|
81 |
-
language:
|
82 |
-
- en
|
83 |
-
pretty_name: GLDv2 Top 51 Categories
|
84 |
-
size_categories:
|
85 |
-
- n<1K
|
|
|
33 |
This dataset is a subset of Kaggle's Google Landmark Recognition 2021 competition with only the categories with more than 500 images.
|
34 |
https://www.kaggle.com/competitions/landmark-recognition-2021/data
|
35 |
|
36 |
+
The dataset consists of a total of 45579 224x224 color images in 51 categories.
|
|
|
|
|
37 |
|
38 |
### Languages
|
39 |
|
40 |
English
|
41 |
|
42 |
+
|
43 |
+
## Dataset Structure
|
44 |
+
|
45 |
+
### Data Fields
|
46 |
+
|
47 |
+
- `landmark_id`: Int - Numeric identifier of the category
|
48 |
+
- `category` : String - Name of the category
|
49 |
+
- `id` : String - Image identifier
|
50 |
+
- `image` : Image - PIL image object
|
51 |
+
- `label` : Int - Numeric label from 0 to 50
|
52 |
+
|
53 |
+
|
54 |
+
### Data Splits
|
55 |
+
|
56 |
+
The dataset was randomly split with 80% of the images for the train set and 20% for the test set.
|
57 |
+
|
58 |
+
| | train | test |
|
59 |
+
|----------------------|------:|-----:|
|
60 |
+
| Dataset | 36463 | 9116 |
|
61 |
+
|
62 |
+
|
63 |
### Source Data
|
64 |
|
65 |
The full dataset is from Kaggle Landmark Recognition 2021
|
|
|
73 |
"Google Landmarks Dataset v2 - A Large-Scale Benchmark for Instance-Level Recognition and Retrieval", T. Weyand, A. Araujo, B. Cao and J. Sim, Proc. CVPR'20
|
74 |
|
75 |
"Towards A Fairer Landmark Recognition Dataset", Z. Kim, A. Araujo, B. Cao, C. Askew, J. Sim, M. Green, N. Yilla and T. Weyand, arxiv:2108.08874
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|