samH98/LungCancerDetection

1import argparse2import time3from pathlib import Path4 5import cv26import torch7import torch.backends.cudnn as cudnn8from numpy import random9 10from models.experimental import attempt_load11from utils.datasets import LoadStreams, LoadImages12from utils.general import check_img_size, check_requirements, check_imshow, non_max_suppression, apply_classifier, \13    scale_coords, xyxy2xywh, strip_optimizer, set_logging, increment_path14from utils.plots import plot_one_box15from utils.torch_utils import select_device, load_classifier, time_synchronized, TracedModel16 17 18def detect(save_img=False):19    source, weights, view_img, save_txt, imgsz, trace = opt.source, opt.weights, opt.view_img, opt.save_txt, opt.img_size, not opt.no_trace20    save_img = not opt.nosave and not source.endswith('.txt')  # save inference images21    webcam = source.isnumeric() or source.endswith('.txt') or source.lower().startswith(22        ('rtsp://', 'rtmp://', 'http://', 'https://'))23 24    # Directories25    save_dir = Path(increment_path(Path(opt.project) / opt.name, exist_ok=opt.exist_ok))  # increment run26    (save_dir / 'labels' if save_txt else save_dir).mkdir(parents=True, exist_ok=True)  # make dir27 28    # Initialize29    set_logging()30    device = select_device(opt.device)31    half = device.type != 'cpu'  # half precision only supported on CUDA32 33    # Load model34    model = attempt_load(weights, map_location=device)  # load FP32 model35    stride = int(model.stride.max())  # model stride36    imgsz = check_img_size(imgsz, s=stride)  # check img_size37 38    if trace:39        model = TracedModel(model, device, opt.img_size)40 41    if half:42        model.half()  # to FP1643 44    # Second-stage classifier45    classify = False46    if classify:47        modelc = load_classifier(name='resnet101', n=2)  # initialize48        modelc.load_state_dict(torch.load('weights/resnet101.pt', map_location=device)['model']).to(device).eval()49 50    # Set Dataloader51    vid_path, vid_writer = None, None52    if webcam:53        view_img = check_imshow()54        cudnn.benchmark = True  # set True to speed up constant image size inference55        dataset = LoadStreams(source, img_size=imgsz, stride=stride)56    else:57        dataset = LoadImages(source, img_size=imgsz, stride=stride)58 59    # Get names and colors60    names = model.module.names if hasattr(model, 'module') else model.names61    colors = [[random.randint(0, 255) for _ in range(3)] for _ in names]62 63    # Run inference64    if device.type != 'cpu':65        model(torch.zeros(1, 3, imgsz, imgsz).to(device).type_as(next(model.parameters())))  # run once66    old_img_w = old_img_h = imgsz67    old_img_b = 168 69    t0 = time.time()70    for path, img, im0s, vid_cap in dataset:71        img = torch.from_numpy(img).to(device)72        img = img.half() if half else img.float()  # uint8 to fp16/3273        img /= 255.0  # 0 - 255 to 0.0 - 1.074        if img.ndimension() == 3:75            img = img.unsqueeze(0)76 77        # Warmup78        if device.type != 'cpu' and (old_img_b != img.shape[0] or old_img_h != img.shape[2] or old_img_w != img.shape[3]):79            old_img_b = img.shape[0]80            old_img_h = img.shape[2]81            old_img_w = img.shape[3]82            for i in range(3):83                model(img, augment=opt.augment)[0]84 85        # Inference86        t1 = time_synchronized()87        with torch.no_grad():   # Calculating gradients would cause a GPU memory leak88            pred = model(img, augment=opt.augment)[0]89        t2 = time_synchronized()90 91        # Apply NMS92        pred = non_max_suppression(pred, opt.conf_thres, opt.iou_thres, classes=opt.classes, agnostic=opt.agnostic_nms)93        t3 = time_synchronized()94 95        # Apply Classifier96        if classify:97            pred = apply_classifier(pred, modelc, img, im0s)98 99        # Process detections100        # detected_classes = []101        # names = ['class1', 'class2', 'class3', 'class 4']  # Replace this list with the actual class names used in detect.py102        for i, det in enumerate(pred):  # detections per image103            if webcam:  # batch_size >= 1104                p, s, im0, frame = path[i], '%g: ' % i, im0s[i].copy(), dataset.count105            else:106                p, s, im0, frame = path, '', im0s, getattr(dataset, 'frame', 0)107 108            p = Path(p)  # to Path109            save_path = str(save_dir / p.name)  # img.jpg110            txt_path = str(save_dir / 'labels' / p.stem) + ('' if dataset.mode == 'image' else f'_{frame}')  # img.txt111            gn = torch.tensor(im0.shape)[[1, 0, 1, 0]]  # normalization gain whwh112            if len(det):113                # Rescale boxes from img_size to im0 size114                det[:, :4] = scale_coords(img.shape[2:], det[:, :4], im0.shape).round()115                116                for c in det[:, -1].unique():117                    n = (det[:, -1] == c).sum()118                    s += f"{n} {names[int(c)]}{'s' * (n > 1)},"119                    # detected_classes.append(names[int(c)])120 121                # Print results122                # for c in det[:, -1].unique():123                #     n = (det[:, -1] == c).sum()  # detections per class124                #     s += f"{n} {names[int(c)]}{'s' * (n > 1)}, "  # add to string125 126                # Write results127                for *xyxy, conf, cls in reversed(det):128                    if save_txt:  # Write to file129                        xywh = (xyxy2xywh(torch.tensor(xyxy).view(1, 4)) / gn).view(-1).tolist()  # normalized xywh130                        line = (cls, *xywh, conf) if opt.save_conf else (cls, *xywh)  # label format131                        with open(txt_path + '.txt', 'a') as f:132                            f.write(('%g ' * len(line)).rstrip() % line + '\n')133 134                    if save_img or view_img:  # Add bbox to image135                        label = f'{names[int(cls)]} {conf:.2f}'136                        plot_one_box(xyxy, im0, label=label, color=colors[int(cls)], line_thickness=1)137 138            # Print time (inference + NMS)139            print(f'{s}Done. ({(1E3 * (t2 - t1)):.1f}ms) Inference, ({(1E3 * (t3 - t2)):.1f}ms) NMS')140 141            # Stream results142            if view_img:143                cv2.imshow(str(p), im0)144                cv2.waitKey(1)  # 1 millisecond145 146            # Save results (image with detections)147            if save_img:148                if dataset.mode == 'image':149                    cv2.imwrite(save_path, im0)150                    print(f" The image with the result is saved in: {save_path}")151                else:  # 'video' or 'stream'152                    if vid_path != save_path:  # new video153                        vid_path = save_path154                        if isinstance(vid_writer, cv2.VideoWriter):155                            vid_writer.release()  # release previous video writer156                        if vid_cap:  # video157                            fps = vid_cap.get(cv2.CAP_PROP_FPS)158                            w = int(vid_cap.get(cv2.CAP_PROP_FRAME_WIDTH))159                            h = int(vid_cap.get(cv2.CAP_PROP_FRAME_HEIGHT))160                        else:  # stream161                            fps, w, h = 30, im0.shape[1], im0.shape[0]162                            save_path += '.mp4'163                        vid_writer = cv2.VideoWriter(save_path, cv2.VideoWriter_fourcc(*'mp4v'), fps, (w, h))164                    vid_writer.write(im0)165 166    if save_txt or save_img:167        s = f"\n{len(list(save_dir.glob('labels/*.txt')))} labels saved to {save_dir / 'labels'}" if save_txt else ''168        #print(f"Results saved to {save_dir}{s}")169 170    print(f'Done. ({time.time() - t0:.3f}s)')171    # return detected_classes172 173 174if __name__ == '__main__':175    parser = argparse.ArgumentParser()176    parser.add_argument('--weights', nargs='+', type=str, default='yolov7.pt', help='model.pt path(s)')177    parser.add_argument('--source', type=str, default='inference/images', help='source')  # file/folder, 0 for webcam178    parser.add_argument('--img-size', type=int, default=640, help='inference size (pixels)')179    parser.add_argument('--conf-thres', type=float, default=0.25, help='object confidence threshold')180    parser.add_argument('--iou-thres', type=float, default=0.45, help='IOU threshold for NMS')181    parser.add_argument('--device', default='', help='cuda device, i.e. 0 or 0,1,2,3 or cpu')182    parser.add_argument('--view-img', action='store_true', help='display results')183    parser.add_argument('--save-txt', action='store_true', help='save results to *.txt')184    parser.add_argument('--save-conf', action='store_true', help='save confidences in --save-txt labels')185    parser.add_argument('--nosave', action='store_true', help='do not save images/videos')186    parser.add_argument('--classes', nargs='+', type=int, help='filter by class: --class 0, or --class 0 2 3')187    parser.add_argument('--agnostic-nms', action='store_true', help='class-agnostic NMS')188    parser.add_argument('--augment', action='store_true', help='augmented inference')189    parser.add_argument('--update', action='store_true', help='update all models')190    parser.add_argument('--project', default='runs/detect', help='save results to project/name')191    parser.add_argument('--name', default='exp', help='save results to project/name')192    parser.add_argument('--exist-ok', action='store_true', help='existing project/name ok, do not increment')193    parser.add_argument('--no-trace', action='store_true', help='don`t trace model')194    opt = parser.parse_args()195    print(opt)196    #check_requirements(exclude=('pycocotools', 'thop'))197 198    with torch.no_grad():199        if opt.update:  # update all models (to fix SourceChangeWarning)200            for opt.weights in ['yolov7.pt']:201                detect()202                strip_optimizer(opt.weights)203        else:204            detect()205