Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -1,64 +1,231 @@
|
|
1 |
"""
|
2 |
-
|
3 |
-
|
|
|
|
|
4 |
"""
|
5 |
|
|
|
|
|
|
|
|
|
6 |
import os
|
|
|
|
|
7 |
from ultralytics import YOLO
|
8 |
|
9 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
10 |
"""
|
11 |
-
|
12 |
-
|
13 |
-
Args:
|
14 |
-
model_name: Name of the YOLOv8 model to download
|
15 |
-
save_dir: Directory to save the model to
|
16 |
"""
|
17 |
-
|
18 |
-
|
19 |
-
|
20 |
-
|
21 |
-
|
22 |
-
|
23 |
-
|
24 |
-
|
25 |
-
|
26 |
-
|
27 |
-
|
28 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
29 |
try:
|
30 |
-
model
|
|
|
|
|
31 |
print("Model loaded successfully")
|
32 |
-
return model_path
|
33 |
except Exception as e:
|
34 |
-
print(f"Error loading
|
35 |
-
print("
|
36 |
-
|
37 |
-
|
38 |
-
|
39 |
-
|
40 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
41 |
|
42 |
-
#
|
43 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
44 |
|
45 |
-
#
|
46 |
-
|
47 |
-
|
48 |
-
import shutil
|
49 |
-
shutil.copy(model_file, model_path)
|
50 |
-
print(f"Model saved to: {model_path}")
|
51 |
|
52 |
-
|
53 |
-
|
54 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
55 |
return None
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
56 |
|
57 |
-
|
58 |
-
|
59 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
60 |
|
61 |
-
|
62 |
-
|
63 |
-
|
64 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
"""
|
2 |
+
Phone Detection App for Hugging Face Spaces
|
3 |
+
|
4 |
+
This app uses YOLOv8 to detect phones in real-time through a webcam feed.
|
5 |
+
When a phone is detected, a warning message is displayed.
|
6 |
"""
|
7 |
|
8 |
+
import cv2
|
9 |
+
import numpy as np
|
10 |
+
import torch
|
11 |
+
import time
|
12 |
import os
|
13 |
+
import gradio as gr
|
14 |
+
from PIL import Image, ImageDraw, ImageFont
|
15 |
from ultralytics import YOLO
|
16 |
|
17 |
+
# Configurations
|
18 |
+
MODEL_PATH = "models/yolov8n.pt" # Path to the model within the repository
|
19 |
+
TARGET_CLASS = "cell phone"
|
20 |
+
TARGET_CLASS_ID = 67 # In YOLOv8's COCO dataset
|
21 |
+
MIN_CONFIDENCE = 0.4 # Minimum confidence threshold for detections
|
22 |
+
|
23 |
+
class PhoneDetector:
|
24 |
"""
|
25 |
+
A class to handle phone detection using YOLOv8 model
|
|
|
|
|
|
|
|
|
26 |
"""
|
27 |
+
def __init__(self, model_path=MODEL_PATH, confidence=MIN_CONFIDENCE):
|
28 |
+
"""
|
29 |
+
Initialize the phone detector
|
30 |
+
|
31 |
+
Args:
|
32 |
+
model_path: Path to the YOLOv8 model weights
|
33 |
+
confidence: Minimum confidence threshold for detections
|
34 |
+
"""
|
35 |
+
self.target_class = TARGET_CLASS
|
36 |
+
self.target_class_id = TARGET_CLASS_ID
|
37 |
+
self.min_confidence = confidence
|
38 |
+
|
39 |
+
# Select device (GPU if available, otherwise CPU)
|
40 |
+
self.device = "cuda" if torch.cuda.is_available() else "cpu"
|
41 |
+
print(f"Using device: {self.device}")
|
42 |
+
|
43 |
+
# Check if model exists, otherwise use default YOLOv8n
|
44 |
+
if not os.path.exists(model_path):
|
45 |
+
print(f"Model not found at {model_path}, using default YOLOv8n")
|
46 |
+
model_path = "yolov8n.pt" # Will be downloaded automatically by YOLO
|
47 |
+
|
48 |
+
# Load model
|
49 |
try:
|
50 |
+
print(f"Loading YOLOv8 model from {model_path}...")
|
51 |
+
self.model = YOLO(model_path)
|
52 |
+
self.model.to(self.device)
|
53 |
print("Model loaded successfully")
|
|
|
54 |
except Exception as e:
|
55 |
+
print(f"Error loading model: {e}")
|
56 |
+
print("Loading default YOLOv8n model...")
|
57 |
+
self.model = YOLO("yolov8n.pt")
|
58 |
+
self.model.to(self.device)
|
59 |
+
|
60 |
+
def detect(self, frame):
|
61 |
+
"""
|
62 |
+
Detect phones in a frame and add visualization
|
63 |
+
|
64 |
+
Args:
|
65 |
+
frame: Input image frame (numpy array)
|
66 |
+
|
67 |
+
Returns:
|
68 |
+
Processed frame with detection visualization
|
69 |
+
"""
|
70 |
+
if frame is None:
|
71 |
+
return None
|
72 |
+
|
73 |
+
# Convert to RGB if grayscale
|
74 |
+
if len(frame.shape) == 2:
|
75 |
+
frame = cv2.cvtColor(frame, cv2.COLOR_GRAY2RGB)
|
76 |
+
elif frame.shape[2] == 4: # If RGBA
|
77 |
+
frame = cv2.cvtColor(frame, cv2.COLOR_RGBA2RGB)
|
78 |
+
|
79 |
+
# Get frame dimensions
|
80 |
+
(h, w) = frame.shape[:2]
|
81 |
+
|
82 |
+
# Convert to PIL Image for easier text rendering
|
83 |
+
pil_image = Image.fromarray(frame)
|
84 |
+
draw = ImageDraw.Draw(pil_image)
|
85 |
+
|
86 |
+
# Try to load a nicer font, fall back to default if not available
|
87 |
+
try:
|
88 |
+
font = ImageFont.truetype("DejaVuSans.ttf", 25)
|
89 |
+
small_font = ImageFont.truetype("DejaVuSans.ttf", 15)
|
90 |
+
except IOError:
|
91 |
+
font = ImageFont.load_default()
|
92 |
+
small_font = ImageFont.load_default()
|
93 |
+
|
94 |
+
# Perform detection with YOLOv8
|
95 |
+
with torch.no_grad(): # Disable gradient calculation for inference
|
96 |
+
results = self.model.predict(frame, conf=self.min_confidence, verbose=False)
|
97 |
+
|
98 |
+
# Flag to track if a phone is detected in this frame
|
99 |
+
phone_detected = False
|
100 |
+
|
101 |
+
# Process detection results
|
102 |
+
if len(results) > 0:
|
103 |
+
for result in results:
|
104 |
+
boxes = result.boxes
|
105 |
+
for box in boxes:
|
106 |
+
# Get class ID
|
107 |
+
cls_id = int(box.cls[0].item())
|
108 |
+
class_name = result.names[cls_id]
|
109 |
+
|
110 |
+
# Check if the detected object is a cell phone
|
111 |
+
if class_name == self.target_class or cls_id == self.target_class_id:
|
112 |
+
phone_detected = True
|
113 |
+
|
114 |
+
# Get confidence score
|
115 |
+
conf = float(box.conf[0].item())
|
116 |
+
|
117 |
+
# Get bounding box coordinates
|
118 |
+
x1, y1, x2, y2 = map(int, box.xyxy[0])
|
119 |
+
|
120 |
+
# Draw bounding box on PIL image
|
121 |
+
draw.rectangle([(x1, y1), (x2, y2)], outline="red", width=3)
|
122 |
+
|
123 |
+
# Display confidence and class
|
124 |
+
label = f"{class_name}: {conf:.2f}"
|
125 |
+
y_label = y1 - 15 if y1 - 15 > 15 else y1 + 15
|
126 |
+
draw.text((x1, y_label), label, fill="red", font=small_font)
|
127 |
|
128 |
+
# Display warning message if phone is detected
|
129 |
+
if phone_detected:
|
130 |
+
warning_text = "WARNING: Phone Detected!"
|
131 |
+
|
132 |
+
# Measure text size for centering (implementation differs based on PIL version)
|
133 |
+
try:
|
134 |
+
# For newer PIL versions
|
135 |
+
text_width = draw.textlength(warning_text, font=font)
|
136 |
+
except AttributeError:
|
137 |
+
# For older PIL versions
|
138 |
+
text_width = font.getmask(warning_text).getbbox()[2]
|
139 |
+
|
140 |
+
text_x = (w - text_width) // 2
|
141 |
+
text_y = h // 2
|
142 |
+
|
143 |
+
# Draw semi-transparent red rectangle for warning
|
144 |
+
overlay = Image.new('RGBA', pil_image.size, (0, 0, 0, 0))
|
145 |
+
overlay_draw = ImageDraw.Draw(overlay)
|
146 |
+
overlay_draw.rectangle([(0, text_y - 40), (w, text_y + 10)], fill=(255, 0, 0, 128))
|
147 |
+
pil_image = Image.alpha_composite(pil_image.convert('RGBA'), overlay).convert('RGB')
|
148 |
+
draw = ImageDraw.Draw(pil_image)
|
149 |
+
|
150 |
+
# Draw warning text
|
151 |
+
draw.text((text_x, text_y - 30), warning_text, fill="white", font=font)
|
152 |
|
153 |
+
# Add processing info at the bottom
|
154 |
+
device_text = f"Running on: {self.device}"
|
155 |
+
draw.text((10, h - 30), device_text, fill="green", font=small_font)
|
|
|
|
|
|
|
156 |
|
157 |
+
# Convert back to numpy array
|
158 |
+
result_frame = np.array(pil_image)
|
159 |
+
|
160 |
+
return result_frame
|
161 |
+
|
162 |
+
# Initialize the detector
|
163 |
+
detector = PhoneDetector()
|
164 |
+
|
165 |
+
# Function to process webcam frames
|
166 |
+
def process_webcam(image):
|
167 |
+
"""
|
168 |
+
Process webcam input for Gradio interface
|
169 |
+
|
170 |
+
Args:
|
171 |
+
image: Input image from Gradio
|
172 |
+
|
173 |
+
Returns:
|
174 |
+
Processed image with phone detection visualization
|
175 |
+
"""
|
176 |
+
if image is None:
|
177 |
return None
|
178 |
+
|
179 |
+
# Process the frame
|
180 |
+
result_frame = detector.detect(image)
|
181 |
+
|
182 |
+
if result_frame is None:
|
183 |
+
return image
|
184 |
+
|
185 |
+
return result_frame
|
186 |
|
187 |
+
# Create Gradio interface
|
188 |
+
title = "Phone Detection with YOLOv8"
|
189 |
+
description = """
|
190 |
+
## Real-time Phone Detection
|
191 |
+
|
192 |
+
This app uses YOLOv8 to detect phones in real-time through your webcam.
|
193 |
+
When a phone is detected, a warning message is displayed.
|
194 |
+
|
195 |
+
### How it works:
|
196 |
+
1. The webcam captures your video feed
|
197 |
+
2. Each frame is analyzed by YOLOv8 to detect phones
|
198 |
+
3. If a phone is detected, a warning message appears
|
199 |
+
|
200 |
+
### Notes:
|
201 |
+
- You may need to give permission for camera access
|
202 |
+
- The app works best with good lighting conditions
|
203 |
+
- The model detects cell phones only
|
204 |
+
"""
|
205 |
+
|
206 |
+
# Create Gradio blocks interface
|
207 |
+
with gr.Blocks(title=title) as demo:
|
208 |
+
gr.Markdown(description)
|
209 |
|
210 |
+
with gr.Row():
|
211 |
+
with gr.Column():
|
212 |
+
# Webcam input with streaming
|
213 |
+
webcam_input = gr.Image(label="Webcam", sources=["webcam"], streaming=True)
|
214 |
+
|
215 |
+
with gr.Column():
|
216 |
+
output_display = gr.Image(label="Detection Result")
|
217 |
+
|
218 |
+
# Stream processing
|
219 |
+
webcam_input.stream(process_webcam, inputs=webcam_input, outputs=output_display)
|
220 |
+
|
221 |
+
gr.Markdown("""
|
222 |
+
### Technical Details
|
223 |
+
- Model: YOLOv8n (optimized for speed)
|
224 |
+
- Target class: "cell phone"
|
225 |
+
- Confidence threshold: 0.4
|
226 |
+
|
227 |
+
This application was developed using Ultralytics YOLOv8, Gradio, and OpenCV.
|
228 |
+
""")
|
229 |
+
|
230 |
+
# Launch the interface
|
231 |
+
demo.launch()
|