Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -2,21 +2,46 @@ import gradio as gr
|
|
2 |
import torch
|
3 |
from PIL import Image
|
4 |
from transformers import ColPaliForRetrieval, ColPaliProcessor
|
|
|
5 |
|
6 |
model_name = "vidore/colpali-v1.3-hf"
|
7 |
model = ColPaliForRetrieval.from_pretrained(model_name, torch_dtype=torch.float32).eval()
|
8 |
processor = ColPaliProcessor.from_pretrained(model_name)
|
9 |
|
10 |
def process_image(image):
|
|
|
|
|
|
|
|
|
11 |
inputs = processor(images=image, return_tensors="pt")
|
|
|
|
|
12 |
with torch.no_grad():
|
13 |
outputs = model(**inputs)
|
14 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
15 |
|
|
|
16 |
demo = gr.Interface(
|
17 |
fn=process_image,
|
18 |
inputs=gr.Image(type="pil"),
|
19 |
-
outputs=
|
|
|
|
|
20 |
)
|
21 |
|
|
|
22 |
demo.launch()
|
|
|
2 |
import torch
|
3 |
from PIL import Image
|
4 |
from transformers import ColPaliForRetrieval, ColPaliProcessor
|
5 |
+
import numpy as np
|
6 |
|
7 |
model_name = "vidore/colpali-v1.3-hf"
|
8 |
model = ColPaliForRetrieval.from_pretrained(model_name, torch_dtype=torch.float32).eval()
|
9 |
processor = ColPaliProcessor.from_pretrained(model_name)
|
10 |
|
11 |
def process_image(image):
|
12 |
+
# Ensure the image is in RGB format
|
13 |
+
image = image.convert('RGB')
|
14 |
+
|
15 |
+
# Process the image
|
16 |
inputs = processor(images=image, return_tensors="pt")
|
17 |
+
|
18 |
+
# Generate embeddings
|
19 |
with torch.no_grad():
|
20 |
outputs = model(**inputs)
|
21 |
+
|
22 |
+
# Extract embeddings and convert to list
|
23 |
+
embeddings = outputs.embeddings.squeeze().cpu().numpy().tolist()
|
24 |
+
|
25 |
+
# Truncate the embeddings for display purposes
|
26 |
+
truncated_embeddings = embeddings[:10] # Show only first 10 values
|
27 |
+
|
28 |
+
# Prepare the output
|
29 |
+
output = {
|
30 |
+
"embedding_sample": truncated_embeddings,
|
31 |
+
"embedding_length": len(embeddings),
|
32 |
+
"embedding_shape": list(np.array(embeddings).shape)
|
33 |
+
}
|
34 |
+
|
35 |
+
return output
|
36 |
|
37 |
+
# Create Gradio interface
|
38 |
demo = gr.Interface(
|
39 |
fn=process_image,
|
40 |
inputs=gr.Image(type="pil"),
|
41 |
+
outputs=gr.JSON(),
|
42 |
+
title="ColPali Image Embedding Generator",
|
43 |
+
description="Upload an image to generate its embedding using the ColPali model."
|
44 |
)
|
45 |
|
46 |
+
# Launch the interface
|
47 |
demo.launch()
|