Spaces:
Running
Running
| import torch | |
| import cv2 | |
| import numpy as np | |
| from PIL import Image | |
| import torchvision.transforms as transforms | |
| from pytorch_grad_cam import EigenCAM | |
| from pytorch_grad_cam.utils.image import show_cam_on_image, scale_cam_image | |
| from ultralytics import YOLO | |
| COLORS = np.random.uniform(0, 255, size=(80, 3)) | |
| def parse_detections_yolov8(results): | |
| boxes, colors, names = [], [], [] | |
| detections = results.boxes | |
| for box in detections: | |
| confidence = box.conf[0].item() | |
| if confidence < 0.2: | |
| continue | |
| xmin, ymin, xmax, ymax = map(int, box.xyxy[0].tolist()) | |
| category = int(box.cls[0].item()) | |
| name = results.names[category] | |
| boxes.append((xmin, ymin, xmax, ymax)) | |
| colors.append(COLORS[category]) | |
| names.append(name) | |
| return boxes, colors, names | |
| def draw_detections(boxes, colors, names, img): | |
| for box, color, name in zip(boxes, colors, names): | |
| xmin, ymin, xmax, ymax = box | |
| cv2.rectangle(img, (xmin, ymin), (xmax, ymax), color, 2) | |
| cv2.putText(img, name, (xmin, ymin - 5), | |
| cv2.FONT_HERSHEY_SIMPLEX, 0.8, color, 2, | |
| lineType=cv2.LINE_AA) | |
| return img | |
| def generate_cam_image(model, target_layers, tensor, rgb_img, boxes): | |
| cam = EigenCAM(model, target_layers) | |
| grayscale_cam = cam(tensor)[0, :, :] | |
| img_float = np.float32(rgb_img) / 255 | |
| # Generate Grad-CAM | |
| cam_image = show_cam_on_image(img_float, grayscale_cam, use_rgb=True) | |
| # Renormalize Grad-CAM inside bounding boxes | |
| renormalized_cam = np.zeros(grayscale_cam.shape, dtype=np.float32) | |
| for x1, y1, x2, y2 in boxes: | |
| renormalized_cam[y1:y2, x1:x2] = scale_cam_image(grayscale_cam[y1:y2, x1:x2].copy()) | |
| renormalized_cam = scale_cam_image(renormalized_cam) | |
| renormalized_cam_image = show_cam_on_image(img_float, renormalized_cam, use_rgb=True) | |
| return cam_image, renormalized_cam_image | |
| def xai_yolov8(image): | |
| # Load YOLOv8 model | |
| model = YOLO('yolov8n.pt') # Load YOLOv8 nano model | |
| model.to('cpu') | |
| model.eval() | |
| # Run YOLO detection | |
| results = model(image) | |
| boxes, colors, names = parse_detections_yolov8(results[0]) | |
| detections_img = draw_detections(boxes, colors, names, image.copy()) | |
| # Prepare input tensor for Grad-CAM | |
| img_float = np.float32(image) / 255 | |
| transform = transforms.ToTensor() | |
| tensor = transform(img_float).unsqueeze(0) | |
| # Grad-CAM visualization | |
| target_layers = [model.model.model[-2]] # Adjust the target layer if required | |
| cam_image, renormalized_cam_image = generate_cam_image(model.model, target_layers, tensor, image, boxes) | |
| # Combine results | |
| final_image = np.hstack((image, cam_image, renormalized_cam_image)) | |
| caption = "Results using YOLOv8" | |
| return Image.fromarray(final_image), caption |