Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -6,14 +6,18 @@ import os
|
|
6 |
import tempfile
|
7 |
import spaces
|
8 |
import gradio as gr
|
9 |
-
|
10 |
import subprocess
|
11 |
import sys
|
|
|
|
|
|
|
|
|
|
|
|
|
12 |
|
13 |
def install_flash_attn_wheel():
|
14 |
flash_attn_wheel_url = "https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.4cxx11abiFALSE-cp310-cp310-linux_x86_64.whl"
|
15 |
try:
|
16 |
-
# Call pip to install the wheel file
|
17 |
subprocess.check_call([sys.executable, "-m", "pip", "install", flash_attn_wheel_url])
|
18 |
print("Wheel installed successfully!")
|
19 |
except subprocess.CalledProcessError as e:
|
@@ -21,7 +25,6 @@ def install_flash_attn_wheel():
|
|
21 |
|
22 |
install_flash_attn_wheel()
|
23 |
|
24 |
-
import cv2
|
25 |
try:
|
26 |
from mmengine.visualization import Visualizer
|
27 |
except ImportError:
|
@@ -43,25 +46,75 @@ tokenizer = AutoTokenizer.from_pretrained(
|
|
43 |
trust_remote_code = True,
|
44 |
)
|
45 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
46 |
from third_parts import VideoReader
|
47 |
def read_video(video_path, video_interval):
|
48 |
vid_frames = VideoReader(video_path)[::video_interval]
|
49 |
|
50 |
temp_dir = tempfile.mkdtemp()
|
51 |
os.makedirs(temp_dir, exist_ok=True)
|
52 |
-
image_paths = []
|
53 |
|
54 |
for frame_idx in range(len(vid_frames)):
|
55 |
frame_image = vid_frames[frame_idx]
|
56 |
-
frame_image = frame_image[..., ::-1]
|
57 |
frame_image = Image.fromarray(frame_image)
|
58 |
vid_frames[frame_idx] = frame_image
|
59 |
-
|
60 |
-
# Save the frame as a .jpg file in the temporary folder
|
61 |
image_path = os.path.join(temp_dir, f"frame_{frame_idx:04d}.jpg")
|
62 |
frame_image.save(image_path, format="JPEG")
|
63 |
-
|
64 |
-
# Append the image path to the list
|
65 |
image_paths.append(image_path)
|
66 |
return vid_frames, image_paths
|
67 |
|
@@ -71,17 +124,10 @@ def visualize(pred_mask, image_path, work_dir):
|
|
71 |
visualizer.set_image(img)
|
72 |
visualizer.draw_binary_masks(pred_mask, colors='g', alphas=0.4)
|
73 |
visual_result = visualizer.get_image()
|
74 |
-
|
75 |
output_path = os.path.join(work_dir, os.path.basename(image_path))
|
76 |
cv2.imwrite(output_path, visual_result)
|
77 |
return output_path
|
78 |
|
79 |
-
|
80 |
-
|
81 |
-
# μ½λ μλ¨μ import μΆκ°
|
82 |
-
from deep_translator import GoogleTranslator
|
83 |
-
|
84 |
-
# λ²μ ν¨μ μμ
|
85 |
def translate_to_korean(text):
|
86 |
try:
|
87 |
translator = GoogleTranslator(source='en', target='ko')
|
@@ -92,7 +138,6 @@ def translate_to_korean(text):
|
|
92 |
|
93 |
@spaces.GPU
|
94 |
def image_vision(image_input_path, prompt):
|
95 |
-
# νκΈ μ
λ ₯ νμΈ
|
96 |
is_korean = any(ord('κ°') <= ord(char) <= ord('ν£') for char in prompt)
|
97 |
|
98 |
image_path = image_input_path
|
@@ -109,9 +154,7 @@ def image_vision(image_input_path, prompt):
|
|
109 |
print(return_dict)
|
110 |
answer = return_dict["prediction"]
|
111 |
|
112 |
-
# νκΈ ν둬ννΈμΈ κ²½μ° μλ΅μ νκΈλ‘ λ²μ
|
113 |
if is_korean:
|
114 |
-
# [SEG]λ 보쑴νλ©΄μ λλ¨Έμ§ ν
μ€νΈλ§ λ²μ
|
115 |
if '[SEG]' in answer:
|
116 |
parts = answer.split('[SEG]')
|
117 |
translated_parts = [translate_to_korean(part.strip()) for part in parts]
|
@@ -133,7 +176,6 @@ def image_vision(image_input_path, prompt):
|
|
133 |
|
134 |
@spaces.GPU(duration=80)
|
135 |
def video_vision(video_input_path, prompt, video_interval):
|
136 |
-
# νκΈ μ
λ ₯ νμΈ
|
137 |
is_korean = any(ord('κ°') <= ord(char) <= ord('ν£') for char in prompt)
|
138 |
|
139 |
cap = cv2.VideoCapture(video_input_path)
|
@@ -151,7 +193,6 @@ def video_vision(video_input_path, prompt, video_interval):
|
|
151 |
prediction = result['prediction']
|
152 |
print(prediction)
|
153 |
|
154 |
-
# νκΈ ν둬ννΈμΈ κ²½μ° μλ΅μ νκΈλ‘ λ²μ
|
155 |
if is_korean:
|
156 |
if '[SEG]' in prediction:
|
157 |
parts = prediction.split('[SEG]')
|
@@ -185,31 +226,38 @@ def video_vision(video_input_path, prompt, video_interval):
|
|
185 |
print(f"Video created successfully at {output_video}")
|
186 |
|
187 |
return prediction, output_video
|
188 |
-
|
189 |
else:
|
190 |
return prediction, None
|
191 |
|
192 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
193 |
|
|
|
194 |
with gr.Blocks(analytics_enabled=False) as demo:
|
195 |
with gr.Column():
|
196 |
gr.Markdown("# Sa2VA: Marrying SAM2 with LLaVA for Dense Grounded Understanding of Images and Videos")
|
197 |
-
|
198 |
-
|
199 |
-
<a href="https://github.com/magic-research/Sa2VA">
|
200 |
-
<img src='https://img.shields.io/badge/GitHub-Repo-blue'>
|
201 |
-
</a>
|
202 |
-
<a href="https://arxiv.org/abs/2501.04001">
|
203 |
-
<img src='https://img.shields.io/badge/ArXiv-Paper-red'>
|
204 |
-
</a>
|
205 |
-
<a href="https://huggingface.co/spaces/fffiloni/Sa2VA-simple-demo?duplicate=true">
|
206 |
-
<img src="https://huggingface.co/datasets/huggingface/badges/resolve/main/duplicate-this-space-sm.svg" alt="Duplicate this Space">
|
207 |
-
</a>
|
208 |
-
<a href="https://huggingface.co/fffiloni">
|
209 |
-
<img src="https://huggingface.co/datasets/huggingface/badges/resolve/main/follow-me-on-HF-sm-dark.svg" alt="Follow me on HF">
|
210 |
-
</a>
|
211 |
-
</div>
|
212 |
-
""")
|
213 |
with gr.Tab("Single Image"):
|
214 |
with gr.Row():
|
215 |
with gr.Column():
|
@@ -226,6 +274,7 @@ with gr.Blocks(analytics_enabled=False) as demo:
|
|
226 |
inputs = [image_input, instruction],
|
227 |
outputs = [output_res, output_image]
|
228 |
)
|
|
|
229 |
with gr.Tab("Video"):
|
230 |
with gr.Row():
|
231 |
with gr.Column():
|
@@ -243,5 +292,34 @@ with gr.Blocks(analytics_enabled=False) as demo:
|
|
243 |
inputs = [video_input, vid_instruction, frame_interval],
|
244 |
outputs = [vid_output_res, output_video]
|
245 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
246 |
|
247 |
demo.queue().launch(show_api=False, show_error=True)
|
|
|
6 |
import tempfile
|
7 |
import spaces
|
8 |
import gradio as gr
|
|
|
9 |
import subprocess
|
10 |
import sys
|
11 |
+
import cv2
|
12 |
+
import threading
|
13 |
+
import queue
|
14 |
+
import time
|
15 |
+
from collections import deque
|
16 |
+
from deep_translator import GoogleTranslator
|
17 |
|
18 |
def install_flash_attn_wheel():
|
19 |
flash_attn_wheel_url = "https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu123torch2.4cxx11abiFALSE-cp310-cp310-linux_x86_64.whl"
|
20 |
try:
|
|
|
21 |
subprocess.check_call([sys.executable, "-m", "pip", "install", flash_attn_wheel_url])
|
22 |
print("Wheel installed successfully!")
|
23 |
except subprocess.CalledProcessError as e:
|
|
|
25 |
|
26 |
install_flash_attn_wheel()
|
27 |
|
|
|
28 |
try:
|
29 |
from mmengine.visualization import Visualizer
|
30 |
except ImportError:
|
|
|
46 |
trust_remote_code = True,
|
47 |
)
|
48 |
|
49 |
+
class WebcamProcessor:
|
50 |
+
def __init__(self, model, tokenizer, fps_target=15, buffer_size=5):
|
51 |
+
self.model = model
|
52 |
+
self.tokenizer = tokenizer
|
53 |
+
self.fps_target = fps_target
|
54 |
+
self.frame_interval = 1.0 / fps_target
|
55 |
+
self.buffer_size = buffer_size
|
56 |
+
self.frame_buffer = deque(maxlen=buffer_size)
|
57 |
+
self.result_queue = queue.Queue()
|
58 |
+
self.is_running = False
|
59 |
+
self.last_process_time = 0
|
60 |
+
|
61 |
+
def start(self):
|
62 |
+
self.is_running = True
|
63 |
+
self.capture = cv2.VideoCapture(0)
|
64 |
+
self.capture_thread = threading.Thread(target=self._capture_loop)
|
65 |
+
self.process_thread = threading.Thread(target=self._process_loop)
|
66 |
+
self.capture_thread.start()
|
67 |
+
self.process_thread.start()
|
68 |
+
|
69 |
+
def stop(self):
|
70 |
+
self.is_running = False
|
71 |
+
if hasattr(self, 'capture_thread'):
|
72 |
+
self.capture_thread.join()
|
73 |
+
self.process_thread.join()
|
74 |
+
self.capture.release()
|
75 |
+
|
76 |
+
def _capture_loop(self):
|
77 |
+
while self.is_running:
|
78 |
+
ret, frame = self.capture.read()
|
79 |
+
if ret:
|
80 |
+
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
|
81 |
+
frame = cv2.resize(frame, (640, 480))
|
82 |
+
current_time = time.time()
|
83 |
+
if current_time - self.last_process_time >= self.frame_interval:
|
84 |
+
self.frame_buffer.append(frame)
|
85 |
+
self.last_process_time = current_time
|
86 |
+
|
87 |
+
def _process_loop(self):
|
88 |
+
while self.is_running:
|
89 |
+
if len(self.frame_buffer) >= self.buffer_size:
|
90 |
+
frames = list(self.frame_buffer)
|
91 |
+
try:
|
92 |
+
result = self.model.predict_forward(
|
93 |
+
video=frames,
|
94 |
+
text="<image>Describe what you see",
|
95 |
+
tokenizer=self.tokenizer
|
96 |
+
)
|
97 |
+
self.result_queue.put(result)
|
98 |
+
except Exception as e:
|
99 |
+
print(f"Processing error: {e}")
|
100 |
+
self.frame_buffer.clear()
|
101 |
+
time.sleep(0.1)
|
102 |
+
|
103 |
from third_parts import VideoReader
|
104 |
def read_video(video_path, video_interval):
|
105 |
vid_frames = VideoReader(video_path)[::video_interval]
|
106 |
|
107 |
temp_dir = tempfile.mkdtemp()
|
108 |
os.makedirs(temp_dir, exist_ok=True)
|
109 |
+
image_paths = []
|
110 |
|
111 |
for frame_idx in range(len(vid_frames)):
|
112 |
frame_image = vid_frames[frame_idx]
|
113 |
+
frame_image = frame_image[..., ::-1]
|
114 |
frame_image = Image.fromarray(frame_image)
|
115 |
vid_frames[frame_idx] = frame_image
|
|
|
|
|
116 |
image_path = os.path.join(temp_dir, f"frame_{frame_idx:04d}.jpg")
|
117 |
frame_image.save(image_path, format="JPEG")
|
|
|
|
|
118 |
image_paths.append(image_path)
|
119 |
return vid_frames, image_paths
|
120 |
|
|
|
124 |
visualizer.set_image(img)
|
125 |
visualizer.draw_binary_masks(pred_mask, colors='g', alphas=0.4)
|
126 |
visual_result = visualizer.get_image()
|
|
|
127 |
output_path = os.path.join(work_dir, os.path.basename(image_path))
|
128 |
cv2.imwrite(output_path, visual_result)
|
129 |
return output_path
|
130 |
|
|
|
|
|
|
|
|
|
|
|
|
|
131 |
def translate_to_korean(text):
|
132 |
try:
|
133 |
translator = GoogleTranslator(source='en', target='ko')
|
|
|
138 |
|
139 |
@spaces.GPU
|
140 |
def image_vision(image_input_path, prompt):
|
|
|
141 |
is_korean = any(ord('κ°') <= ord(char) <= ord('ν£') for char in prompt)
|
142 |
|
143 |
image_path = image_input_path
|
|
|
154 |
print(return_dict)
|
155 |
answer = return_dict["prediction"]
|
156 |
|
|
|
157 |
if is_korean:
|
|
|
158 |
if '[SEG]' in answer:
|
159 |
parts = answer.split('[SEG]')
|
160 |
translated_parts = [translate_to_korean(part.strip()) for part in parts]
|
|
|
176 |
|
177 |
@spaces.GPU(duration=80)
|
178 |
def video_vision(video_input_path, prompt, video_interval):
|
|
|
179 |
is_korean = any(ord('κ°') <= ord(char) <= ord('ν£') for char in prompt)
|
180 |
|
181 |
cap = cv2.VideoCapture(video_input_path)
|
|
|
193 |
prediction = result['prediction']
|
194 |
print(prediction)
|
195 |
|
|
|
196 |
if is_korean:
|
197 |
if '[SEG]' in prediction:
|
198 |
parts = prediction.split('[SEG]')
|
|
|
226 |
print(f"Video created successfully at {output_video}")
|
227 |
|
228 |
return prediction, output_video
|
|
|
229 |
else:
|
230 |
return prediction, None
|
231 |
|
232 |
+
@spaces.GPU
|
233 |
+
def webcam_vision(prompt):
|
234 |
+
is_korean = any(ord('κ°') <= ord(char) <= ord('ν£') for char in prompt)
|
235 |
+
|
236 |
+
if not hasattr(webcam_vision, 'processor'):
|
237 |
+
webcam_vision.processor = WebcamProcessor(model, tokenizer)
|
238 |
+
|
239 |
+
if not webcam_vision.processor.is_running:
|
240 |
+
webcam_vision.processor.start()
|
241 |
+
|
242 |
+
try:
|
243 |
+
result = webcam_vision.processor.result_queue.get(timeout=5)
|
244 |
+
prediction = result['prediction']
|
245 |
+
|
246 |
+
if is_korean:
|
247 |
+
prediction = translate_to_korean(prediction)
|
248 |
+
|
249 |
+
return prediction
|
250 |
+
except queue.Empty:
|
251 |
+
return "No results available yet"
|
252 |
+
except Exception as e:
|
253 |
+
return f"Error: {str(e)}"
|
254 |
|
255 |
+
# Gradio UI
|
256 |
with gr.Blocks(analytics_enabled=False) as demo:
|
257 |
with gr.Column():
|
258 |
gr.Markdown("# Sa2VA: Marrying SAM2 with LLaVA for Dense Grounded Understanding of Images and Videos")
|
259 |
+
|
260 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
261 |
with gr.Tab("Single Image"):
|
262 |
with gr.Row():
|
263 |
with gr.Column():
|
|
|
274 |
inputs = [image_input, instruction],
|
275 |
outputs = [output_res, output_image]
|
276 |
)
|
277 |
+
|
278 |
with gr.Tab("Video"):
|
279 |
with gr.Row():
|
280 |
with gr.Column():
|
|
|
292 |
inputs = [video_input, vid_instruction, frame_interval],
|
293 |
outputs = [vid_output_res, output_video]
|
294 |
)
|
295 |
+
|
296 |
+
with gr.Tab("Webcam"):
|
297 |
+
with gr.Row():
|
298 |
+
with gr.Column():
|
299 |
+
webcam_input = gr.Image(source="webcam", streaming=True)
|
300 |
+
with gr.Row():
|
301 |
+
webcam_instruction = gr.Textbox(
|
302 |
+
label="Instruction",
|
303 |
+
placeholder="Enter instruction here...",
|
304 |
+
scale=4
|
305 |
+
)
|
306 |
+
start_button = gr.Button("Start", scale=1)
|
307 |
+
stop_button = gr.Button("Stop", scale=1)
|
308 |
+
with gr.Column():
|
309 |
+
webcam_output = gr.Textbox(label="Response")
|
310 |
+
processed_view = gr.Image(label="Processed View")
|
311 |
+
|
312 |
+
status_text = gr.Textbox(label="Status", value="Ready")
|
313 |
+
|
314 |
+
start_button.click(
|
315 |
+
fn=lambda x: webcam_vision(x),
|
316 |
+
inputs=[webcam_instruction],
|
317 |
+
outputs=[webcam_output]
|
318 |
+
)
|
319 |
+
|
320 |
+
stop_button.click(
|
321 |
+
fn=lambda: "Stopped" if hasattr(webcam_vision, 'processor') and webcam_vision.processor.stop() else "Not running",
|
322 |
+
outputs=[status_text]
|
323 |
+
)
|
324 |
|
325 |
demo.queue().launch(show_api=False, show_error=True)
|