Spaces:
Running
on
Zero
Running
on
Zero
Update app.py
Browse files
app.py
CHANGED
@@ -1,32 +1,61 @@
|
|
1 |
-
import PIL.Image as Image
|
2 |
import gradio as gr
|
3 |
-
import
|
4 |
-
|
5 |
from ultralytics import YOLOv10
|
|
|
6 |
|
7 |
@spaces.GPU
|
8 |
-
def
|
9 |
model = YOLOv10.from_pretrained(f'jameslahm/{model_id}')
|
10 |
-
|
11 |
-
source=
|
12 |
-
|
13 |
-
|
14 |
-
|
15 |
-
|
16 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
17 |
|
18 |
-
|
19 |
-
|
20 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
21 |
|
22 |
-
return im
|
23 |
|
24 |
def app():
|
25 |
with gr.Blocks():
|
26 |
with gr.Row():
|
27 |
with gr.Column():
|
28 |
-
image = gr.Image(type="pil", label="Image")
|
29 |
-
|
|
|
|
|
|
|
|
|
|
|
30 |
model_id = gr.Dropdown(
|
31 |
label="Model",
|
32 |
choices=[
|
@@ -56,35 +85,52 @@ def app():
|
|
56 |
yolov10_infer = gr.Button(value="Detect Objects")
|
57 |
|
58 |
with gr.Column():
|
59 |
-
output_image = gr.Image(type="
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
60 |
|
61 |
yolov10_infer.click(
|
62 |
-
fn=
|
63 |
-
inputs=[
|
64 |
-
|
65 |
-
model_id,
|
66 |
-
image_size,
|
67 |
-
conf_threshold,
|
68 |
-
],
|
69 |
-
outputs=[output_image],
|
70 |
)
|
71 |
|
72 |
gr.Examples(
|
73 |
examples=[
|
74 |
[
|
75 |
-
"bus.jpg",
|
76 |
"yolov10s",
|
77 |
640,
|
78 |
0.25,
|
79 |
],
|
80 |
[
|
81 |
-
"zidane.jpg",
|
82 |
"yolov10s",
|
83 |
640,
|
84 |
0.25,
|
85 |
],
|
86 |
],
|
87 |
-
fn=
|
88 |
inputs=[
|
89 |
image,
|
90 |
model_id,
|
@@ -113,4 +159,4 @@ with gradio_app:
|
|
113 |
with gr.Column():
|
114 |
app()
|
115 |
if __name__ == '__main__':
|
116 |
-
gradio_app.launch()
|
|
|
|
|
1 |
import gradio as gr
|
2 |
+
import cv2
|
3 |
+
import tempfile
|
4 |
from ultralytics import YOLOv10
|
5 |
+
import spaces
|
6 |
|
7 |
@spaces.GPU
|
8 |
+
def yolov10_inference(image, video, model_id, image_size, conf_threshold):
|
9 |
model = YOLOv10.from_pretrained(f'jameslahm/{model_id}')
|
10 |
+
if image:
|
11 |
+
results = model.predict(source=image, imgsz=image_size, conf=conf_threshold)
|
12 |
+
annotated_image = results[0].plot()
|
13 |
+
return annotated_image[:, :, ::-1], None
|
14 |
+
else:
|
15 |
+
video_path = tempfile.mktemp(suffix=".webm")
|
16 |
+
with open(video_path, "wb") as f:
|
17 |
+
with open(video, "rb") as g:
|
18 |
+
f.write(g.read())
|
19 |
+
|
20 |
+
cap = cv2.VideoCapture(video_path)
|
21 |
+
fps = cap.get(cv2.CAP_PROP_FPS)
|
22 |
+
frame_width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
|
23 |
+
frame_height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
|
24 |
+
|
25 |
+
output_video_path = tempfile.mktemp(suffix=".webm")
|
26 |
+
out = cv2.VideoWriter(output_video_path, cv2.VideoWriter_fourcc(*'vp80'), fps, (frame_width, frame_height))
|
27 |
+
|
28 |
+
while cap.isOpened():
|
29 |
+
ret, frame = cap.read()
|
30 |
+
if not ret:
|
31 |
+
break
|
32 |
|
33 |
+
results = model.predict(source=frame, imgsz=image_size, conf=conf_threshold)
|
34 |
+
annotated_frame = results[0].plot()
|
35 |
+
out.write(annotated_frame)
|
36 |
+
|
37 |
+
cap.release()
|
38 |
+
out.release()
|
39 |
+
|
40 |
+
return None, output_video_path
|
41 |
+
|
42 |
+
@spaces.GPU
|
43 |
+
def yolov10_inference_for_examples(image, model_path, image_size, conf_threshold):
|
44 |
+
annotated_image, _ = yolov10_inference(image, None, model_path, image_size, conf_threshold)
|
45 |
+
return annotated_image
|
46 |
|
|
|
47 |
|
48 |
def app():
|
49 |
with gr.Blocks():
|
50 |
with gr.Row():
|
51 |
with gr.Column():
|
52 |
+
image = gr.Image(type="pil", label="Image", visible=True)
|
53 |
+
video = gr.Video(label="Video", visible=False)
|
54 |
+
input_type = gr.Radio(
|
55 |
+
choices=["Image", "Video"],
|
56 |
+
value="Image",
|
57 |
+
label="Input Type",
|
58 |
+
)
|
59 |
model_id = gr.Dropdown(
|
60 |
label="Model",
|
61 |
choices=[
|
|
|
85 |
yolov10_infer = gr.Button(value="Detect Objects")
|
86 |
|
87 |
with gr.Column():
|
88 |
+
output_image = gr.Image(type="numpy", label="Annotated Image", visible=True)
|
89 |
+
output_video = gr.Video(label="Annotated Video", visible=False)
|
90 |
+
|
91 |
+
def update_visibility(input_type):
|
92 |
+
image = gr.update(visible=True) if input_type == "Image" else gr.update(visible=False)
|
93 |
+
video = gr.update(visible=False) if input_type == "Image" else gr.update(visible=True)
|
94 |
+
output_image = gr.update(visible=True) if input_type == "Image" else gr.update(visible=False)
|
95 |
+
output_video = gr.update(visible=False) if input_type == "Image" else gr.update(visible=True)
|
96 |
+
|
97 |
+
return image, video, output_image, output_video
|
98 |
+
|
99 |
+
input_type.change(
|
100 |
+
fn=update_visibility,
|
101 |
+
inputs=[input_type],
|
102 |
+
outputs=[image, video, output_image, output_video],
|
103 |
+
)
|
104 |
+
|
105 |
+
def run_inference(image, video, model_id, image_size, conf_threshold, input_type):
|
106 |
+
if input_type == "Image":
|
107 |
+
return yolov10_inference(image, None, model_id, image_size, conf_threshold)
|
108 |
+
else:
|
109 |
+
return yolov10_inference(None, video, model_id, image_size, conf_threshold)
|
110 |
+
|
111 |
|
112 |
yolov10_infer.click(
|
113 |
+
fn=run_inference,
|
114 |
+
inputs=[image, video, model_id, image_size, conf_threshold, input_type],
|
115 |
+
outputs=[output_image, output_video],
|
|
|
|
|
|
|
|
|
|
|
116 |
)
|
117 |
|
118 |
gr.Examples(
|
119 |
examples=[
|
120 |
[
|
121 |
+
"ultralytics/assets/bus.jpg",
|
122 |
"yolov10s",
|
123 |
640,
|
124 |
0.25,
|
125 |
],
|
126 |
[
|
127 |
+
"ultralytics/assets/zidane.jpg",
|
128 |
"yolov10s",
|
129 |
640,
|
130 |
0.25,
|
131 |
],
|
132 |
],
|
133 |
+
fn=yolov10_inference_for_examples,
|
134 |
inputs=[
|
135 |
image,
|
136 |
model_id,
|
|
|
159 |
with gr.Column():
|
160 |
app()
|
161 |
if __name__ == '__main__':
|
162 |
+
gradio_app.launch()
|