PythonでYOLOv5のresultを取得したい

#PythonでYOLOv5のresultを取得したい

前提・実現したいこと

コンソール上にしか結果が表示されないのでそれをtxtファイルに出力したいです。

発生している問題・エラーメッセージ

yolov5-masterというディレクトリにYOLOv5のファイルを解凍してあります。
その中のdetect.pyをpython detect.py --source 0で実行しました。
そうするとコマンドプロントには下記のものはその一部ですが結果が返ってきます。
ですがこれをtxtファイルに出力できなくて困っています。

試したこと

とりあえずprintしているところが118行目の

python
1print(f'{s}Done. ({t2 - t1:.3f}s)')

だと分かったのでこれを変数に格納して
強引に結果を切り取ったりしてみました。

python
1result = f'{s}Done. ({t2 - t1:.3f}s)'
2result = result[14:]

Code

Python
1detect.py
2
3import argparse
4import time
5from pathlib import Path
6
7import cv2
8import torch
9import torch.backends.cudnn as cudnn
10from numpy import random
11
12from models.experimental import attempt_load
13from utils.datasets import LoadStreams, LoadImages
14from utils.general import check_img_size, check_requirements, check_imshow, non_max_suppression, apply_classifier, \
15    scale_coords, xyxy2xywh, strip_optimizer, set_logging, increment_path
16from utils.plots import plot_one_box
17from utils.torch_utils import select_device, load_classifier, time_synchronized
18
19
20def detect(save_img=False):
21    source, weights, view_img, save_txt, imgsz = opt.source, opt.weights, opt.view_img, opt.save_txt, opt.img_size
22    webcam = source.isnumeric() or source.endswith('.txt') or source.lower().startswith(
23        ('rtsp://', 'rtmp://', 'http://'))
24
25    # Directories
26    save_dir = Path(increment_path(Path(opt.project) / opt.name, exist_ok=opt.exist_ok))  # increment run
27    (save_dir / 'labels' if save_txt else save_dir).mkdir(parents=True, exist_ok=True)  # make dir
28
29    # Initialize
30    set_logging()
31    device = select_device(opt.device)
32    half = device.type != 'cpu'  # half precision only supported on CUDA
33
34    # Load model
35    model = attempt_load(weights, map_location=device)  # load FP32 model
36    stride = int(model.stride.max())  # model stride
37    imgsz = check_img_size(imgsz, s=stride)  # check img_size
38    if half:
39        model.half()  # to FP16
40
41    # Second-stage classifier
42    classify = False
43    if classify:
44        modelc = load_classifier(name='resnet101', n=2)  # initialize
45        modelc.load_state_dict(torch.load('weights/resnet101.pt', map_location=device)['model']).to(device).eval()
46
47    # Set Dataloader
48    vid_path, vid_writer = None, None
49    if webcam:
50        view_img = check_imshow()
51        cudnn.benchmark = True  # set True to speed up constant image size inference
52        dataset = LoadStreams(source, img_size=imgsz, stride=stride)
53    else:
54        save_img = True
55        dataset = LoadImages(source, img_size=imgsz, stride=stride)
56
57    # Get names and colors
58    names = model.module.names if hasattr(model, 'module') else model.names
59    colors = [[random.randint(0, 255) for _ in range(3)] for _ in names]
60
61    # Run inference
62    if device.type != 'cpu':
63        model(torch.zeros(1, 3, imgsz, imgsz).to(device).type_as(next(model.parameters())))  # run once
64    t0 = time.time()
65    for path, img, im0s, vid_cap in dataset:
66        img = torch.from_numpy(img).to(device)
67        img = img.half() if half else img.float()  # uint8 to fp16/32
68        img /= 255.0  # 0 - 255 to 0.0 - 1.0
69        if img.ndimension() == 3:
70            img = img.unsqueeze(0)
71
72        # Inference
73        t1 = time_synchronized()
74        pred = model(img, augment=opt.augment)[0]
75
76        # Apply NMS
77        pred = non_max_suppression(pred, opt.conf_thres, opt.iou_thres, classes=opt.classes, agnostic=opt.agnostic_nms)
78        t2 = time_synchronized()
79
80        # Apply Classifier
81        if classify:
82            pred = apply_classifier(pred, modelc, img, im0s)
83
84        # Process detections
85        for i, det in enumerate(pred):  # detections per image
86            if webcam:  # batch_size >= 1
87                p, s, im0, frame = path[i], '%g: ' % i, im0s[i].copy(), dataset.count
88            else:
89                p, s, im0, frame = path, '', im0s, getattr(dataset, 'frame', 0)
90
91            p = Path(p)  # to Path
92            save_path = str(save_dir / p.name)  # img.jpg
93            txt_path = str(save_dir / 'labels' / p.stem) + ('' if dataset.mode == 'image' else f'_{frame}')  # img.txt
94            s += '%gx%g ' % img.shape[2:]  # print string
95            gn = torch.tensor(im0.shape)[[1, 0, 1, 0]]  # normalization gain whwh
96            if len(det):
97                # Rescale boxes from img_size to im0 size
98                det[:, :4] = scale_coords(img.shape[2:], det[:, :4], im0.shape).round()
99
100                # Print results
101                for c in det[:, -1].unique():
102                    n = (det[:, -1] == c).sum()  # detections per class
103                    s += f"{n} {names[int(c)]}{'s' * (n > 1)}, "  # add to string
104
105                # Write results
106                for *xyxy, conf, cls in reversed(det):
107                    if save_txt:  # Write to file
108                        xywh = (xyxy2xywh(torch.tensor(xyxy).view(1, 4)) / gn).view(-1).tolist()  # normalized xywh
109                        line = (cls, *xywh, conf) if opt.save_conf else (cls, *xywh)  # label format
110                        with open(txt_path + '.txt', 'a') as f:
111                            f.write(('%g ' * len(line)).rstrip() % line + '\n')
112
113                    if save_img or view_img:  # Add bbox to image
114                        label = f'{names[int(cls)]} {conf:.2f}'
115                        plot_one_box(xyxy, im0, label=label, color=colors[int(cls)], line_thickness=3)
116
117            # Print time (inference + NMS)
118            print(f'{s}Done. ({t2 - t1:.3f}s)')
119
120            # Stream results
121            if view_img:
122                cv2.imshow(str(p), im0)
123                cv2.waitKey(1)  # 1 millisecond
124
125            # Save results (image with detections)
126            if save_img:
127                if dataset.mode == 'image':
128                    cv2.imwrite(save_path, im0)
129                else:  # 'video'
130                    if vid_path != save_path:  # new video
131                        vid_path = save_path
132                        if isinstance(vid_writer, cv2.VideoWriter):
133                            vid_writer.release()  # release previous video writer
134
135                        fourcc = 'mp4v'  # output video codec
136                        fps = vid_cap.get(cv2.CAP_PROP_FPS)
137                        w = int(vid_cap.get(cv2.CAP_PROP_FRAME_WIDTH))
138                        h = int(vid_cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
139                        vid_writer = cv2.VideoWriter(save_path, cv2.VideoWriter_fourcc(*fourcc), fps, (w, h))
140                    vid_writer.write(im0)
141
142    if save_txt or save_img:
143        s = f"\n{len(list(save_dir.glob('labels/*.txt')))} labels saved to {save_dir / 'labels'}" if save_txt else ''
144        print(f"Results saved to {save_dir}{s}")
145
146    print(f'Done. ({time.time() - t0:.3f}s)')
147
148
149if __name__ == '__main__':
150    parser = argparse.ArgumentParser()
151    parser.add_argument('--weights', nargs='+', type=str, default='yolov5s.pt', help='model.pt path(s)')
152    parser.add_argument('--source', type=str, default='data/images', help='source')  # file/folder, 0 for webcam
153    parser.add_argument('--img-size', type=int, default=640, help='inference size (pixels)')
154    parser.add_argument('--conf-thres', type=float, default=0.25, help='object confidence threshold')
155    parser.add_argument('--iou-thres', type=float, default=0.45, help='IOU threshold for NMS')
156    parser.add_argument('--device', default='', help='cuda device, i.e. 0 or 0,1,2,3 or cpu')
157    parser.add_argument('--view-img', action='store_true', help='display results')
158    parser.add_argument('--save-txt', action='store_true', help='save results to *.txt')
159    parser.add_argument('--save-conf', action='store_true', help='save confidences in --save-txt labels')
160    parser.add_argument('--classes', nargs='+', type=int, help='filter by class: --class 0, or --class 0 2 3')
161    parser.add_argument('--agnostic-nms', action='store_true', help='class-agnostic NMS')
162    parser.add_argument('--augment', action='store_true', help='augmented inference')
163    parser.add_argument('--update', action='store_true', help='update all models')
164    parser.add_argument('--project', default='runs/detect', help='save results to project/name')
165    parser.add_argument('--name', default='exp', help='save results to project/name')
166    parser.add_argument('--exist-ok', action='store_true', help='existing project/name ok, do not increment')
167    opt = parser.parse_args()
168    print(opt)
169    check_requirements()
170
171    with torch.no_grad():
172        if opt.update:  # update all models (to fix SourceChangeWarning)
173            for opt.weights in ['yolov5s.pt', 'yolov5m.pt', 'yolov5l.pt', 'yolov5x.pt']:
174                detect()
175                strip_optimizer(opt.weights)
176        else:
177            detect()

cmd
10: 480x640 1 tv, Done. (0.219s)
20: 480x640 1 tv, 1 book, Done. (0.224s)
30: 480x640 1 person, Done. (0.215s)
40: 480x640 Done. (0.217s)
50: 480x640 Done. (0.212s)
60: 480x640 1 tv, Done. (0.205s)
70: 480x640 1 tie, Done. (0.208s)
80: 480x640 Done. (0.211s)
90: 480x640 1 person, Done. (0.219s)
100: 480x640 Done. (0.204s)
110: 480x640 1 chair, Done. (0.218s)
120: 480x640 1 carrot, Done. (0.210s)
130: 480x640 Done. (0.212s)
140: 480x640 Done. (0.215s)
150: 480x640 1 person, Done. (0.211s)
160: 480x640 1 person, Done. (0.214s)
170: 480x640 1 person, 1 tv, 1 laptop, Done. (0.208s)