Update streamlit_app.py
Browse files- streamlit_app.py +66 -217
streamlit_app.py
CHANGED
@@ -1,11 +1,13 @@
|
|
1 |
import asyncio
|
2 |
import sys
|
3 |
|
|
|
4 |
if sys.platform.startswith('linux') and sys.version_info >= (3, 8):
|
5 |
try:
|
6 |
asyncio.set_event_loop_policy(asyncio.DefaultEventLoopPolicy())
|
7 |
except Exception:
|
8 |
pass
|
|
|
9 |
import streamlit as st
|
10 |
from PIL import Image
|
11 |
import numpy as np
|
@@ -15,7 +17,9 @@ import tempfile
|
|
15 |
import os
|
16 |
from ultralytics import YOLO
|
17 |
import cv2 as cv
|
18 |
-
|
|
|
|
|
19 |
|
20 |
model_path="best.pt"
|
21 |
|
@@ -31,19 +35,17 @@ st.set_page_config(
|
|
31 |
st.sidebar.title("🚗 Driver Distraction System")
|
32 |
st.sidebar.write("Choose an option below:")
|
33 |
|
34 |
-
# Sidebar navigation
|
35 |
page = st.sidebar.radio("Select Feature", [
|
36 |
-
"Distraction System",
|
37 |
-
"
|
38 |
-
"
|
39 |
])
|
40 |
|
41 |
# --- Class Labels (for YOLO model) ---
|
|
|
42 |
class_names = ['drinking', 'hair and makeup', 'operating the radio', 'reaching behind',
|
43 |
'safe driving', 'talking on the phone', 'talking to passenger', 'texting']
|
44 |
-
|
45 |
-
# Sidebar Class Name Display
|
46 |
-
st.sidebar.subheader("Class Names")
|
47 |
for idx, class_name in enumerate(class_names):
|
48 |
st.sidebar.write(f"{idx}: {class_name}")
|
49 |
|
@@ -86,233 +88,80 @@ if page == "Distraction System":
|
|
86 |
else:
|
87 |
st.warning("No distractions detected.")
|
88 |
|
89 |
-
else: # Video processing
|
90 |
-
uploaded_video = st.file_uploader("Upload Video", type=["mp4", "avi", "mov", "mkv", "webm"])
|
91 |
-
|
92 |
-
if uploaded_video is not None:
|
93 |
-
tfile = tempfile.NamedTemporaryFile(delete=False, suffix=".mp4")
|
94 |
-
tfile.write(uploaded_video.read())
|
95 |
-
temp_input_path = tfile.name
|
96 |
-
temp_output_path = tempfile.mktemp(suffix="_distraction_detected.mp4")
|
97 |
-
|
98 |
-
st.subheader("Video Information")
|
99 |
-
cap = cv.VideoCapture(temp_input_path)
|
100 |
-
fps = cap.get(cv.CAP_PROP_FPS)
|
101 |
-
width = int(cap.get(cv.CAP_PROP_FRAME_WIDTH))
|
102 |
-
height = int(cap.get(cv.CAP_PROP_FRAME_HEIGHT))
|
103 |
-
total_frames = int(cap.get(cv.CAP_PROP_FRAME_COUNT))
|
104 |
-
duration = total_frames / fps if fps > 0 else 0
|
105 |
-
cap.release()
|
106 |
-
|
107 |
-
col1, col2 = st.columns(2)
|
108 |
-
with col1:
|
109 |
-
st.metric("Duration", f"{duration:.2f} seconds")
|
110 |
-
st.metric("Original FPS", f"{fps:.2f}")
|
111 |
-
with col2:
|
112 |
-
st.metric("Resolution", f"{width}x{height}")
|
113 |
-
st.metric("Total Frames", total_frames)
|
114 |
-
|
115 |
-
st.subheader("Original Video Preview")
|
116 |
-
st.video(uploaded_video)
|
117 |
-
|
118 |
-
if st.button("Process Video for Distraction Detection"):
|
119 |
-
TARGET_PROCESSING_FPS = 10
|
120 |
-
# --- NEW: Hyperparameter for the temporal smoothing logic ---
|
121 |
-
PERSISTENCE_CONFIDENCE_THRESHOLD = 0.40 # Stick with old class if found with >= 40% confidence
|
122 |
-
|
123 |
-
st.info(f"🚀 For faster results, video will be processed at ~{TARGET_PROCESSING_FPS} FPS.")
|
124 |
-
st.info(f"🧠Applying temporal smoothing to reduce status flickering (Persistence Threshold: {PERSISTENCE_CONFIDENCE_THRESHOLD*100:.0f}%).")
|
125 |
-
|
126 |
-
progress_bar = st.progress(0, text="Starting video processing...")
|
127 |
-
|
128 |
-
with st.spinner(f"Processing video... This may take a while."):
|
129 |
-
model = YOLO(model_path)
|
130 |
-
cap = cv.VideoCapture(temp_input_path)
|
131 |
-
|
132 |
-
fourcc = cv.VideoWriter_fourcc(*'mp4v')
|
133 |
-
out = cv.VideoWriter(temp_output_path, fourcc, fps, (width, height))
|
134 |
-
|
135 |
-
frame_skip_interval = max(1, round(fps / TARGET_PROCESSING_FPS))
|
136 |
-
|
137 |
-
frame_count = 0
|
138 |
-
last_best_box_coords = None
|
139 |
-
last_best_box_label = ""
|
140 |
-
last_status_text = "Status: Initializing..."
|
141 |
-
last_status_color = (128, 128, 128)
|
142 |
-
# --- NEW: State variable to store the last confirmed class ---
|
143 |
-
last_confirmed_class_name = 'safe driving'
|
144 |
-
|
145 |
-
while cap.isOpened():
|
146 |
-
ret, frame = cap.read()
|
147 |
-
if not ret:
|
148 |
-
break
|
149 |
-
|
150 |
-
frame_count += 1
|
151 |
-
progress = int((frame_count / total_frames) * 100) if total_frames > 0 else 0
|
152 |
-
progress_bar.progress(progress, text=f"Analyzing frame {frame_count}/{total_frames}")
|
153 |
-
|
154 |
-
annotated_frame = frame.copy()
|
155 |
-
|
156 |
-
if frame_count % frame_skip_interval == 0:
|
157 |
-
results = model(annotated_frame)
|
158 |
-
result = results[0]
|
159 |
-
|
160 |
-
last_best_box_coords = None # Reset box for this processing cycle
|
161 |
-
|
162 |
-
if len(result.boxes) > 0:
|
163 |
-
boxes = result.boxes
|
164 |
-
class_names_dict = result.names
|
165 |
-
confidences = boxes.conf.cpu().numpy()
|
166 |
-
classes = boxes.cls.cpu().numpy()
|
167 |
-
|
168 |
-
# --- NEW STABILITY LOGIC ---
|
169 |
-
final_box_to_use = None
|
170 |
-
|
171 |
-
# 1. Check if the last known class exists with reasonable confidence
|
172 |
-
for i in range(len(boxes)):
|
173 |
-
current_class_name = class_names_dict[int(classes[i])]
|
174 |
-
if current_class_name == last_confirmed_class_name and confidences[i] >= PERSISTENCE_CONFIDENCE_THRESHOLD:
|
175 |
-
final_box_to_use = boxes[i]
|
176 |
-
break
|
177 |
-
|
178 |
-
# 2. If not, fall back to the highest confidence detection in the current frame
|
179 |
-
if final_box_to_use is None:
|
180 |
-
max_conf_idx = confidences.argmax()
|
181 |
-
final_box_to_use = boxes[max_conf_idx]
|
182 |
-
# --- END OF NEW LOGIC ---
|
183 |
-
|
184 |
-
# Now, process the determined "final_box_to_use"
|
185 |
-
x1, y1, x2, y2 = final_box_to_use.xyxy[0].cpu().numpy()
|
186 |
-
confidence = final_box_to_use.conf[0].cpu().numpy()
|
187 |
-
class_id = int(final_box_to_use.cls[0].cpu().numpy())
|
188 |
-
class_name = class_names_dict[class_id]
|
189 |
-
|
190 |
-
# Update the state for the next frames
|
191 |
-
last_confirmed_class_name = class_name
|
192 |
-
last_best_box_coords = (int(x1), int(y1), int(x2), int(y2))
|
193 |
-
last_best_box_label = f"{class_name}: {confidence:.2f}"
|
194 |
-
|
195 |
-
if class_name != 'safe driving':
|
196 |
-
last_status_text = f"Status: {class_name.replace('_', ' ').title()}"
|
197 |
-
last_status_color = (0, 0, 255)
|
198 |
-
else:
|
199 |
-
last_status_text = "Status: Safe Driving"
|
200 |
-
last_status_color = (0, 128, 0)
|
201 |
-
else:
|
202 |
-
# No detections, reset to safe driving
|
203 |
-
last_confirmed_class_name = 'safe driving'
|
204 |
-
last_status_text = "Status: Safe Driving"
|
205 |
-
last_status_color = (0, 128, 0)
|
206 |
-
|
207 |
-
# Draw annotations on EVERY frame using the last known data
|
208 |
-
if last_best_box_coords:
|
209 |
-
cv.rectangle(annotated_frame, (last_best_box_coords[0], last_best_box_coords[1]),
|
210 |
-
(last_best_box_coords[2], last_best_box_coords[3]), (0, 255, 0), 2)
|
211 |
-
cv.putText(annotated_frame, last_best_box_label,
|
212 |
-
(last_best_box_coords[0], last_best_box_coords[1] - 10),
|
213 |
-
cv.FONT_HERSHEY_SIMPLEX, 0.9, (0, 255, 0), 2)
|
214 |
-
|
215 |
-
# Draw status text
|
216 |
-
font_scale, font_thickness = 1.0, 2
|
217 |
-
(text_w, text_h), _ = cv.getTextSize(last_status_text, cv.FONT_HERSHEY_SIMPLEX, font_scale, font_thickness)
|
218 |
-
padding = 10
|
219 |
-
rect_start = (padding, padding)
|
220 |
-
rect_end = (padding + text_w + padding, padding + text_h + padding)
|
221 |
-
cv.rectangle(annotated_frame, rect_start, rect_end, last_status_color, -1)
|
222 |
-
text_pos = (padding + 5, padding + text_h + 5)
|
223 |
-
cv.putText(annotated_frame, last_status_text, text_pos, cv.FONT_HERSHEY_SIMPLEX, font_scale, (255, 255, 255), font_thickness)
|
224 |
-
|
225 |
-
out.write(annotated_frame)
|
226 |
-
|
227 |
-
cap.release()
|
228 |
-
out.release()
|
229 |
-
progress_bar.progress(100, text="Video processing completed!")
|
230 |
-
|
231 |
-
st.success("Video processed successfully!")
|
232 |
-
|
233 |
-
if os.path.exists(temp_output_path):
|
234 |
-
with open(temp_output_path, "rb") as file:
|
235 |
-
video_bytes = file.read()
|
236 |
-
|
237 |
-
st.download_button(
|
238 |
-
label="📥 Download Processed Video",
|
239 |
-
data=video_bytes,
|
240 |
-
file_name=f"distraction_detected_{uploaded_video.name}",
|
241 |
-
mime="video/mp4",
|
242 |
-
key="download_distraction_video"
|
243 |
-
)
|
244 |
-
|
245 |
-
st.subheader("Sample Frame from Processed Video")
|
246 |
-
cap_out = cv.VideoCapture(temp_output_path)
|
247 |
-
ret, frame = cap_out.read()
|
248 |
-
if ret:
|
249 |
-
frame_rgb = cv.cvtColor(frame, cv.COLOR_BGR2RGB)
|
250 |
-
st.image(frame_rgb, caption="Sample frame with distraction detection", use_container_width=True)
|
251 |
-
cap_out.release()
|
252 |
-
|
253 |
-
try:
|
254 |
-
os.unlink(temp_input_path)
|
255 |
-
if os.path.exists(temp_output_path): os.unlink(temp_output_path)
|
256 |
-
except Exception as e:
|
257 |
-
st.warning(f"Failed to clean up temporary files: {e}")
|
258 |
-
|
259 |
# --- Feature: Real-time Drowsiness Detection ---
|
260 |
elif page == "Real-time Drowsiness Detection":
|
261 |
st.title("🧠Real-time Drowsiness Detection")
|
262 |
-
st.
|
|
|
263 |
if st.button("Start Drowsiness Detection"):
|
264 |
-
|
|
|
265 |
subprocess.Popen(["python3", "drowsiness_detection.py", "--mode", "webcam"])
|
266 |
-
|
|
|
|
|
267 |
|
268 |
# --- Feature: Video Drowsiness Detection ---
|
269 |
elif page == "Video Drowsiness Detection":
|
270 |
st.title("📹 Video Drowsiness Detection")
|
271 |
-
st.write("Upload a video file to detect drowsiness and
|
272 |
uploaded_video = st.file_uploader("Upload Video", type=["mp4", "avi", "mov", "mkv", "webm"])
|
|
|
273 |
if uploaded_video is not None:
|
|
|
274 |
tfile = tempfile.NamedTemporaryFile(delete=False, suffix=".mp4")
|
275 |
tfile.write(uploaded_video.read())
|
276 |
temp_input_path = tfile.name
|
277 |
temp_output_path = tempfile.mktemp(suffix="_processed.mp4")
|
|
|
278 |
st.subheader("Original Video Preview")
|
279 |
st.video(uploaded_video)
|
|
|
280 |
if st.button("Process Video for Drowsiness Detection"):
|
281 |
progress_bar = st.progress(0, text="Preparing to process video...")
|
282 |
-
|
283 |
-
|
284 |
-
|
285 |
-
|
286 |
-
|
287 |
-
"
|
288 |
-
|
289 |
-
stdout, stderr = process.communicate()
|
290 |
-
if process.returncode == 0:
|
291 |
-
progress_bar.progress(100, text="Video processing completed!")
|
292 |
-
if os.path.exists(temp_output_path):
|
293 |
-
st.success("Video processed successfully!")
|
294 |
-
if stdout: st.code(stdout)
|
295 |
-
with open(temp_output_path, "rb") as file: video_bytes = file.read()
|
296 |
-
st.download_button(
|
297 |
-
label="📥 Download Processed Video",
|
298 |
-
data=video_bytes,
|
299 |
-
file_name=f"drowsiness_detected_{uploaded_video.name}",
|
300 |
-
mime="video/mp4",
|
301 |
-
key="download_processed_video"
|
302 |
-
)
|
303 |
-
st.subheader("Sample Frame from Processed Video")
|
304 |
-
cap = cv.VideoCapture(temp_output_path)
|
305 |
-
ret, frame = cap.read()
|
306 |
-
if ret: st.image(cv.cvtColor(frame, cv.COLOR_BGR2RGB), caption="Sample frame with drowsiness detection", use_container_width=True)
|
307 |
-
cap.release()
|
308 |
-
else:
|
309 |
-
st.error("Error: Processed video file not found.")
|
310 |
-
if stderr: st.code(stderr)
|
311 |
-
else:
|
312 |
-
st.error("An error occurred during video processing.")
|
313 |
-
if stderr: st.code(stderr)
|
314 |
try:
|
315 |
-
|
316 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
317 |
except Exception as e:
|
318 |
-
st.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
import asyncio
|
2 |
import sys
|
3 |
|
4 |
+
# --- Boilerplate for compatibility ---
|
5 |
if sys.platform.startswith('linux') and sys.version_info >= (3, 8):
|
6 |
try:
|
7 |
asyncio.set_event_loop_policy(asyncio.DefaultEventLoopPolicy())
|
8 |
except Exception:
|
9 |
pass
|
10 |
+
|
11 |
import streamlit as st
|
12 |
from PIL import Image
|
13 |
import numpy as np
|
|
|
17 |
import os
|
18 |
from ultralytics import YOLO
|
19 |
import cv2 as cv
|
20 |
+
|
21 |
+
# --- NEW: Import your refactored video processing logic ---
|
22 |
+
from video_processor import process_video_with_progress
|
23 |
|
24 |
model_path="best.pt"
|
25 |
|
|
|
35 |
st.sidebar.title("🚗 Driver Distraction System")
|
36 |
st.sidebar.write("Choose an option below:")
|
37 |
|
38 |
+
# --- Sidebar navigation ---
|
39 |
page = st.sidebar.radio("Select Feature", [
|
40 |
+
"Distraction System",
|
41 |
+
"Video Drowsiness Detection",
|
42 |
+
"Real-time Drowsiness Detection"
|
43 |
])
|
44 |
|
45 |
# --- Class Labels (for YOLO model) ---
|
46 |
+
st.sidebar.subheader("Class Names")
|
47 |
class_names = ['drinking', 'hair and makeup', 'operating the radio', 'reaching behind',
|
48 |
'safe driving', 'talking on the phone', 'talking to passenger', 'texting']
|
|
|
|
|
|
|
49 |
for idx, class_name in enumerate(class_names):
|
50 |
st.sidebar.write(f"{idx}: {class_name}")
|
51 |
|
|
|
88 |
else:
|
89 |
st.warning("No distractions detected.")
|
90 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
91 |
# --- Feature: Real-time Drowsiness Detection ---
|
92 |
elif page == "Real-time Drowsiness Detection":
|
93 |
st.title("🧠Real-time Drowsiness Detection")
|
94 |
+
st.info("This feature requires a local webcam and will open a new window.")
|
95 |
+
st.warning("This feature is intended for local use and will not function in the cloud deployment.")
|
96 |
if st.button("Start Drowsiness Detection"):
|
97 |
+
try:
|
98 |
+
# This call is fine, as your new drowsiness_detection.py is set up to handle it.
|
99 |
subprocess.Popen(["python3", "drowsiness_detection.py", "--mode", "webcam"])
|
100 |
+
st.success("Attempted to launch detection window. Please check your desktop.")
|
101 |
+
except Exception as e:
|
102 |
+
st.error(f"Failed to start process: {e}")
|
103 |
|
104 |
# --- Feature: Video Drowsiness Detection ---
|
105 |
elif page == "Video Drowsiness Detection":
|
106 |
st.title("📹 Video Drowsiness Detection")
|
107 |
+
st.write("Upload a video file to detect drowsiness and generate a report.")
|
108 |
uploaded_video = st.file_uploader("Upload Video", type=["mp4", "avi", "mov", "mkv", "webm"])
|
109 |
+
|
110 |
if uploaded_video is not None:
|
111 |
+
# Create a temporary file to hold the uploaded video
|
112 |
tfile = tempfile.NamedTemporaryFile(delete=False, suffix=".mp4")
|
113 |
tfile.write(uploaded_video.read())
|
114 |
temp_input_path = tfile.name
|
115 |
temp_output_path = tempfile.mktemp(suffix="_processed.mp4")
|
116 |
+
|
117 |
st.subheader("Original Video Preview")
|
118 |
st.video(uploaded_video)
|
119 |
+
|
120 |
if st.button("Process Video for Drowsiness Detection"):
|
121 |
progress_bar = st.progress(0, text="Preparing to process video...")
|
122 |
+
|
123 |
+
# --- NEW: Define a callback function for the progress bar ---
|
124 |
+
def streamlit_progress_callback(current, total):
|
125 |
+
if total > 0:
|
126 |
+
percent_complete = int((current / total) * 100)
|
127 |
+
progress_bar.progress(percent_complete, text=f"Analyzing frame {current}/{total}...")
|
128 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
129 |
try:
|
130 |
+
with st.spinner("Processing video... This may take a while."):
|
131 |
+
# --- NEW: Directly call your robust video processing function ---
|
132 |
+
# No more complex subprocess logic needed!
|
133 |
+
stats = process_video_with_progress(
|
134 |
+
input_path=temp_input_path,
|
135 |
+
output_path=temp_output_path,
|
136 |
+
progress_callback=streamlit_progress_callback
|
137 |
+
)
|
138 |
+
|
139 |
+
progress_bar.progress(100, text="Video processing completed!")
|
140 |
+
st.success("Video processed successfully!")
|
141 |
+
|
142 |
+
# --- NEW: Display the returned statistics ---
|
143 |
+
st.subheader("Detection Results")
|
144 |
+
col1, col2, col3 = st.columns(3)
|
145 |
+
col1.metric("Drowsy Events", stats.get('drowsy_events', 0))
|
146 |
+
col2.metric("Yawn Events", stats.get('yawn_events', 0))
|
147 |
+
col3.metric("Head Down Events", stats.get('head_down_events', 0))
|
148 |
+
|
149 |
+
# Offer the processed video for download
|
150 |
+
if os.path.exists(temp_output_path):
|
151 |
+
with open(temp_output_path, "rb") as file:
|
152 |
+
video_bytes = file.read()
|
153 |
+
st.download_button(
|
154 |
+
label="📥 Download Processed Video",
|
155 |
+
data=video_bytes,
|
156 |
+
file_name=f"drowsiness_detected_{uploaded_video.name}",
|
157 |
+
mime="video/mp4"
|
158 |
+
)
|
159 |
except Exception as e:
|
160 |
+
st.error(f"An error occurred during video processing: {e}")
|
161 |
+
finally:
|
162 |
+
# Cleanup temporary files
|
163 |
+
try:
|
164 |
+
if os.path.exists(temp_input_path): os.unlink(temp_input_path)
|
165 |
+
if os.path.exists(temp_output_path): os.unlink(temp_output_path)
|
166 |
+
except Exception as e_clean:
|
167 |
+
st.warning(f"Failed to clean up temporary files: {e_clean}")
|