Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -2,28 +2,43 @@ import gradio as gr
|
|
2 |
import torch
|
3 |
import cv2
|
4 |
import numpy as np
|
5 |
-
from
|
|
|
6 |
|
7 |
-
#
|
8 |
-
|
|
|
|
|
|
|
|
|
9 |
|
10 |
def segment_image(input_image, points):
|
11 |
-
#
|
12 |
-
input_image =
|
|
|
|
|
|
|
13 |
|
14 |
-
#
|
15 |
-
|
|
|
16 |
|
17 |
-
#
|
18 |
-
|
|
|
|
|
|
|
|
|
|
|
19 |
|
20 |
-
#
|
21 |
-
|
22 |
|
23 |
# Overlay the mask on the original image
|
24 |
-
result_image =
|
25 |
-
|
26 |
-
|
|
|
27 |
|
28 |
return result_image
|
29 |
|
@@ -35,8 +50,8 @@ iface = gr.Interface(
|
|
35 |
gr.Image(type="numpy", tool="sketch", brush_radius=5, label="Click on objects to segment")
|
36 |
],
|
37 |
outputs=gr.Image(type="numpy"),
|
38 |
-
title="
|
39 |
-
description="Click on objects in the image to segment them using
|
40 |
)
|
41 |
|
42 |
# Launch the interface
|
|
|
2 |
import torch
|
3 |
import cv2
|
4 |
import numpy as np
|
5 |
+
from transformers import SamModel, SamProcessor
|
6 |
+
from PIL import Image
|
7 |
|
8 |
+
# Set up device
|
9 |
+
device = "cuda" if torch.cuda.is_available() else "cpu"
|
10 |
+
|
11 |
+
# Load model and processor
|
12 |
+
model = SamModel.from_pretrained("facebook/sam-vit-base").to(device)
|
13 |
+
processor = SamProcessor.from_pretrained("facebook/sam-vit-base")
|
14 |
|
15 |
def segment_image(input_image, points):
|
16 |
+
# Convert input_image to PIL Image
|
17 |
+
input_image = Image.fromarray(input_image)
|
18 |
+
|
19 |
+
# Prepare inputs
|
20 |
+
inputs = processor(input_image, input_points=[points], return_tensors="pt").to(device)
|
21 |
|
22 |
+
# Generate masks
|
23 |
+
with torch.no_grad():
|
24 |
+
outputs = model(**inputs)
|
25 |
|
26 |
+
# Post-process masks
|
27 |
+
masks = processor.image_processor.post_process_masks(
|
28 |
+
outputs.pred_masks.cpu(),
|
29 |
+
inputs["original_sizes"].cpu(),
|
30 |
+
inputs["reshaped_input_sizes"].cpu()
|
31 |
+
)
|
32 |
+
scores = outputs.iou_scores
|
33 |
|
34 |
+
# Convert mask to numpy array
|
35 |
+
mask = masks[0][0].numpy()
|
36 |
|
37 |
# Overlay the mask on the original image
|
38 |
+
result_image = np.array(input_image)
|
39 |
+
mask_rgb = np.zeros_like(result_image)
|
40 |
+
mask_rgb[mask > 0.5] = [255, 0, 0] # Red color for the mask
|
41 |
+
result_image = cv2.addWeighted(result_image, 1, mask_rgb, 0.5, 0)
|
42 |
|
43 |
return result_image
|
44 |
|
|
|
50 |
gr.Image(type="numpy", tool="sketch", brush_radius=5, label="Click on objects to segment")
|
51 |
],
|
52 |
outputs=gr.Image(type="numpy"),
|
53 |
+
title="Segment Anything Model (SAM) Image Segmentation",
|
54 |
+
description="Click on objects in the image to segment them using SAM."
|
55 |
)
|
56 |
|
57 |
# Launch the interface
|