Spaces:
Runtime error
Runtime error
File size: 11,335 Bytes
83aae9f |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 |
import cv2
import numpy as np
import time
import os
from ultralytics import YOLO
from PIL import Image
import matplotlib.pyplot as plt
def detect_and_blur(input_source, model=None, frame_skip=3):
"""Detect and blur sensitive elements in images or video frames
Args:
input_source: Image path or video frame (numpy array)
model: YOLO model instance
frame_skip: Frame skip rate for video processing
Returns:
result_rgb: Processed image with blurred regions
detections: Dict with counts of detected objects
boxes: Dict with bounding boxes of detected objects
"""
if isinstance(input_source, str): # Image path
frame = cv2.imread(input_source)
if frame is None:
raise ValueError(f"Could not read image from {input_source}")
else: # Video frame or numpy array
frame = input_source.copy()
# Handle RGB vs BGR input
if len(frame.shape) == 3 and frame.shape[2] == 3:
if isinstance(input_source, str) or input_source is not None:
frame_rgb = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
else:
frame_rgb = frame # Assume RGB if directly passed
else:
raise ValueError("Input must be a color image with 3 channels")
result_img = frame.copy()
detections = {'faces': 0, 'plates': 0}
boxes = {'faces': [], 'plates': []}
if model:
try:
# Run YOLOv8 inference
results = model.predict(frame_rgb, conf=0.5)
for r in results:
for box in r.boxes:
x1, y1, x2, y2 = map(int, box.xyxy[0].tolist())
cls_id = int(box.cls[0])
conf = float(box.conf[0])
# Ensure coordinates are within image bounds
x1, y1 = max(0, x1), max(0, y1)
x2, y2 = min(frame.shape[1], x2), min(frame.shape[0], y2)
# Skip invalid boxes
if x2 <= x1 or y2 <= y1:
continue
# Apply Gaussian blur to the detected region
region = result_img[y1:y2, x1:x2]
# Adjust kernel size based on detection type and region size
kernel_size = 55 if cls_id == 1 else 71 # Different blur for faces vs plates/text
kernel_size = max(25, min(kernel_size, (x2-x1)//2*2+1, (y2-y1)//2*2+1))
kernel_size = kernel_size + 1 if kernel_size % 2 == 0 else kernel_size
# Apply blur only if kernel size is valid
if kernel_size >= 3:
blurred = cv2.GaussianBlur(region, (kernel_size, kernel_size), 15)
result_img[y1:y2, x1:x2] = blurred
# Update detection counts and boxes
if cls_id == 0: # Assuming 0 is plate/text
detections['plates'] += 1
boxes['plates'].append((x1, y1, x2, y2))
else: # Assuming 1 is face
detections['faces'] += 1
boxes['faces'].append((x1, y1, x2, y2))
except Exception as e:
print(f"Detection error: {e}")
# Ensure output is RGB for consistent interface
if isinstance(input_source, str) or input_source is not None:
result_rgb = cv2.cvtColor(result_img, cv2.COLOR_BGR2RGB)
else:
result_rgb = result_img
return result_rgb, detections, boxes
def process_video(input_path, output_path=None, model=None, frame_skip=3, output_fps=30):
"""Process video with optimized frame skipping
Args:
input_path: Path to input video
output_path: Path to save processed video (if None, auto-generated)
model: YOLO model instance
frame_skip: Process 1 in every N frames
output_fps: Output video frame rate
Returns:
output_path: Path to processed video file
"""
try:
# Open video file
cap = cv2.VideoCapture(input_path)
if not cap.isOpened():
raise ValueError(f"Could not open video file {input_path}")
# Get video properties
frame_width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
frame_height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
original_fps = cap.get(cv2.CAP_PROP_FPS)
total_frames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
# Set output parameters
fps = original_fps if original_fps > 0 else output_fps
if output_path is None:
# Create results directory if it doesn't exist
results_dir = os.path.join(os.path.dirname(os.path.abspath(__file__)), "..", "results")
os.makedirs(results_dir, exist_ok=True)
output_path = os.path.join(results_dir, f"processed_{os.path.basename(input_path)}")
# Create video writer
fourcc = cv2.VideoWriter_fourcc(*'mp4v')
out = cv2.VideoWriter(output_path, fourcc, fps//frame_skip, (frame_width, frame_height))
# Display processing information
print(f"Processing video: {os.path.basename(input_path)}")
print(f"Original: {frame_width}x{frame_height} @ {original_fps:.1f}fps")
print(f"Processing: 1 every {frame_skip} frames")
print(f"Output: {fps//frame_skip:.1f}fps | Estimated time: {total_frames/(fps*frame_skip):.1f}s")
# Process video frames
frame_count = 0
processed_frames = 0
start_time = time.time()
while True:
ret, frame = cap.read()
if not ret:
break
# Skip frames according to frame_skip
if frame_count % frame_skip != 0:
frame_count += 1
continue
try:
# Process frame
result_rgb, _, _ = detect_and_blur(frame, model)
result_bgr = cv2.cvtColor(result_rgb, cv2.COLOR_RGB2BGR)
out.write(result_bgr)
processed_frames += 1
# Print progress periodically
if time.time() - start_time >= 5:
elapsed = time.time() - start_time
fps = processed_frames / elapsed
print(f"Progress: {frame_count}/{total_frames} | "
f"Processed: {processed_frames} | "
f"Current FPS: {fps:.1f}")
start_time = time.time()
processed_frames = 0
except Exception as e:
print(f"Error processing frame {frame_count}: {e}")
frame_count += 1
# Clean up
cap.release()
out.release()
print(f"\nVideo processing complete! Saved to {output_path}")
return output_path
except Exception as e:
print(f"Video processing failed: {e}")
return None
def process_image(image_path, output_path=None, model=None, visualize=False):
"""Process single image with optional visualization
Args:
image_path: Path to input image or numpy array
output_path: Path to save processed image (if None, auto-generated)
model: YOLO model instance
visualize: Whether to create visualization with original, detections, and result
Returns:
result_path: Path to processed image file
or
result_rgb: Processed image as numpy array (if output_path is None)
"""
try:
# Process image
result_rgb, detections, boxes = detect_and_blur(image_path, model)
# Handle input as numpy array
if not isinstance(image_path, str):
if output_path is None:
return result_rgb
image_filename = "processed_image.jpg"
else:
image_filename = os.path.basename(image_path)
# Create visualization if requested
if visualize:
# Load original image
if isinstance(image_path, str):
original = cv2.cvtColor(cv2.imread(image_path), cv2.COLOR_BGR2RGB)
else:
original = image_path
# Create image with detection boxes
detection_img = original.copy()
# Draw face boxes
for box in boxes['faces']:
x1, y1, x2, y2 = box
cv2.rectangle(detection_img, (x1, y1), (x2, y2), (255, 0, 0), 2)
cv2.putText(detection_img, "Face", (x1, y1-10),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (255, 0, 0), 2)
# Draw plate/text boxes
for box in boxes['plates']:
x1, y1, x2, y2 = box
cv2.rectangle(detection_img, (x1, y1), (x2, y2), (0, 0, 255), 2)
cv2.putText(detection_img, "Plate/Text", (x1, y1-10),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 0, 255), 2)
# Create comparison visualization (only for local debugging)
fig, axes = plt.subplots(1, 3, figsize=(20, 7))
titles = ["Original", "Detections", "Blurred Result"]
images = [original, detection_img, result_rgb]
for ax, title, img in zip(axes, titles, images):
ax.imshow(img)
ax.set_title(title)
ax.axis("off")
plt.tight_layout()
plt.show()
# Save result if output path provided
if output_path is not None:
# Save the processed image
cv2.imwrite(output_path, cv2.cvtColor(result_rgb, cv2.COLOR_RGB2BGR))
print(f"Saved result to {output_path}")
else:
# Auto-generate output path if not provided
results_dir = os.path.join(os.path.dirname(os.path.abspath(__file__)), "..", "results")
os.makedirs(results_dir, exist_ok=True)
result_path = os.path.join(results_dir, f"processed_{image_filename}")
cv2.imwrite(result_path, cv2.cvtColor(result_rgb, cv2.COLOR_RGB2BGR))
print(f"Saved result to {result_path}")
output_path = result_path
print(f"Detections: {detections['faces']} faces, {detections['plates']} plates/text regions")
return output_path if isinstance(image_path, str) else result_rgb
except Exception as e:
print(f"Image processing error: {e}")
return None
def process_pil_image(pil_image, model=None):
"""Process PIL Image for Gradio interface
Args:
pil_image: PIL Image
model: YOLO model instance
Returns:
result_pil: Processed PIL Image
detections: Dict with counts of detected objects
"""
try:
# Convert PIL to numpy array
img_array = np.array(pil_image)
# Process the image
result_rgb, detections, _ = detect_and_blur(img_array, model)
# Convert back to PIL if needed
result_pil = Image.fromarray(result_rgb)
return result_pil, detections
except Exception as e:
print(f"PIL image processing error: {e}")
return pil_image, {'faces': 0, 'plates': 0} |