Commit
·
ee88db6
1
Parent(s):
cf31499
update
Browse files- app.py +20 -10
- requirements.txt +2 -1
app.py
CHANGED
@@ -6,6 +6,7 @@ import torch
|
|
6 |
import plotly.graph_objects as go
|
7 |
from PIL import Image
|
8 |
import spaces
|
|
|
9 |
|
10 |
@spaces.GPU
|
11 |
def process_images(image1, image2):
|
@@ -17,9 +18,15 @@ def process_images(image1, image2):
|
|
17 |
|
18 |
images = [image1, image2]
|
19 |
processor = AutoImageProcessor.from_pretrained("ETH-CVG/lightglue_superpoint")
|
20 |
-
model = AutoModel.from_pretrained("ETH-CVG/lightglue_superpoint")
|
21 |
-
|
22 |
inputs = processor(images, return_tensors="pt")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
23 |
with torch.no_grad():
|
24 |
outputs = model(**inputs)
|
25 |
|
@@ -39,9 +46,11 @@ def process_images(image1, image2):
|
|
39 |
pil_img = Image.fromarray((image1 / 255.0 * 255).astype(np.uint8))
|
40 |
pil_img2 = Image.fromarray((image2 / 255.0 * 255).astype(np.uint8))
|
41 |
|
42 |
-
# Create Plotly figure
|
43 |
fig = go.Figure()
|
44 |
|
|
|
|
|
|
|
45 |
# Get keypoints
|
46 |
keypoints0_x, keypoints0_y = output["keypoints0"].unbind(1)
|
47 |
keypoints1_x, keypoints1_y = output["keypoints1"].unbind(1)
|
@@ -55,10 +64,13 @@ def process_images(image1, image2):
|
|
55 |
output["matching_scores"],
|
56 |
):
|
57 |
color_val = matching_score.item()
|
58 |
-
|
|
|
|
|
|
|
59 |
|
60 |
hover_text = (
|
61 |
-
f"Score: {matching_score.item():.
|
62 |
f"Point 1: ({keypoint0_x.item():.1f}, {keypoint0_y.item():.1f})<br>"
|
63 |
f"Point 2: ({keypoint1_x.item():.1f}, {keypoint1_y.item():.1f})"
|
64 |
)
|
@@ -78,7 +90,6 @@ def process_images(image1, image2):
|
|
78 |
|
79 |
# Update layout to use images as background
|
80 |
fig.update_layout(
|
81 |
-
title="LightGlue Keypoint Matching",
|
82 |
xaxis=dict(
|
83 |
range=[0, width0 + width1],
|
84 |
showgrid=False,
|
@@ -93,9 +104,8 @@ def process_images(image1, image2):
|
|
93 |
scaleanchor="x",
|
94 |
scaleratio=1,
|
95 |
),
|
96 |
-
margin=dict(l=0, r=0, t=
|
97 |
-
|
98 |
-
width=width0 + width1,
|
99 |
images=[
|
100 |
dict(
|
101 |
source=pil_img,
|
@@ -152,7 +162,7 @@ with gr.Blocks(title="LightGlue Matching Demo") as demo:
|
|
152 |
process_btn = gr.Button("Match Images", variant="primary")
|
153 |
|
154 |
# Output plot
|
155 |
-
output_plot = gr.Plot(label="Matching Results")
|
156 |
|
157 |
# Connect the function
|
158 |
process_btn.click(fn=process_images, inputs=[image1, image2], outputs=output_plot)
|
|
|
6 |
import plotly.graph_objects as go
|
7 |
from PIL import Image
|
8 |
import spaces
|
9 |
+
import matplotlib.cm as cm
|
10 |
|
11 |
@spaces.GPU
|
12 |
def process_images(image1, image2):
|
|
|
18 |
|
19 |
images = [image1, image2]
|
20 |
processor = AutoImageProcessor.from_pretrained("ETH-CVG/lightglue_superpoint")
|
21 |
+
model = AutoModel.from_pretrained("ETH-CVG/lightglue_superpoint", device_map="auto")
|
|
|
22 |
inputs = processor(images, return_tensors="pt")
|
23 |
+
inputs = inputs.to(model.device)
|
24 |
+
print(
|
25 |
+
"Model is on device: ",
|
26 |
+
model.device,
|
27 |
+
"and inputs are on device: ",
|
28 |
+
inputs.device,
|
29 |
+
)
|
30 |
with torch.no_grad():
|
31 |
outputs = model(**inputs)
|
32 |
|
|
|
46 |
pil_img = Image.fromarray((image1 / 255.0 * 255).astype(np.uint8))
|
47 |
pil_img2 = Image.fromarray((image2 / 255.0 * 255).astype(np.uint8))
|
48 |
|
|
|
49 |
fig = go.Figure()
|
50 |
|
51 |
+
# Create colormap (red-yellow-green: red for low scores, green for high scores)
|
52 |
+
colormap = cm.RdYlGn
|
53 |
+
|
54 |
# Get keypoints
|
55 |
keypoints0_x, keypoints0_y = output["keypoints0"].unbind(1)
|
56 |
keypoints1_x, keypoints1_y = output["keypoints1"].unbind(1)
|
|
|
64 |
output["matching_scores"],
|
65 |
):
|
66 |
color_val = matching_score.item()
|
67 |
+
rgba_color = colormap(color_val)
|
68 |
+
|
69 |
+
# Convert to rgba string with transparency
|
70 |
+
color = f"rgba({int(rgba_color[0] * 255)}, {int(rgba_color[1] * 255)}, {int(rgba_color[2] * 255)}, 0.8)"
|
71 |
|
72 |
hover_text = (
|
73 |
+
f"Score: {matching_score.item():.3f}<br>"
|
74 |
f"Point 1: ({keypoint0_x.item():.1f}, {keypoint0_y.item():.1f})<br>"
|
75 |
f"Point 2: ({keypoint1_x.item():.1f}, {keypoint1_y.item():.1f})"
|
76 |
)
|
|
|
90 |
|
91 |
# Update layout to use images as background
|
92 |
fig.update_layout(
|
|
|
93 |
xaxis=dict(
|
94 |
range=[0, width0 + width1],
|
95 |
showgrid=False,
|
|
|
104 |
scaleanchor="x",
|
105 |
scaleratio=1,
|
106 |
),
|
107 |
+
margin=dict(l=0, r=0, t=0, b=0),
|
108 |
+
autosize=True,
|
|
|
109 |
images=[
|
110 |
dict(
|
111 |
source=pil_img,
|
|
|
162 |
process_btn = gr.Button("Match Images", variant="primary")
|
163 |
|
164 |
# Output plot
|
165 |
+
output_plot = gr.Plot(label="Matching Results", scale=2)
|
166 |
|
167 |
# Connect the function
|
168 |
process_btn.click(fn=process_images, inputs=[image1, image2], outputs=output_plot)
|
requirements.txt
CHANGED
@@ -5,4 +5,5 @@ transformers @ git+https://github.com/huggingface/transformers.git@e5a9ce48f711b
|
|
5 |
matplotlib
|
6 |
torch
|
7 |
plotly
|
8 |
-
spaces
|
|
|
|
5 |
matplotlib
|
6 |
torch
|
7 |
plotly
|
8 |
+
spaces
|
9 |
+
accelerate
|