YOLOv7 training and inference

환경 setting

$ conda create -n yolov7 python=3.8
$ conda activate yolov7
$ pip install torch==1.8.1+cu111 torchvision==0.9.1+cu111 torchaudio==0.8.1 -f https://download.pytorch.org/whl/torch_stable.html
$ pip install -r requirements.txt
$ sudo apt-get install libgl1-mesa-glx

or

environment_setting.sh 파일 생성 (in Kubeflow)

#!/bin/sh

echo "################################# Creating conda environment #################################"
conda create -y -n test python=3.8
echo "################################# Activating conda environment #################################"
source /opt/conda/etc/profile.d/conda.sh
source /opt/conda/bin/activate test
echo "################################# Installing pytorch #################################"
pip install torch==1.8.1+cu111 torchvision==0.9.1+cu111 torchaudio==0.8.1 -f https://download.pytorch.org/whl/torch_stable.html
echo "################################# Installing dependencies #################################"
pip install -r requirements.txt
echo "################################# Installing libgl1-mesa-glx #################################"
sudo apt-get install libgl1-mesa-glx
echo "################################# Finished #################################"

environment_setting.sh 파일 실행 (in Kubeflow)

$ . ./environment_setting.sh

Inference

Single Image

$ python detect.py --weights yolov7.pt --conf 0.25 --img-size 640 --source inference/images/myimage.jpg

Dataset

$ python detect.py --weights yolov7.pt --conf 0.25 --img-size 640 --source /path/to/my/dataset

Video

# Merge frames to make video

$ python detect.py --weights yolov7.pt --conf 0.25 --img-size 640 --source /path/to/myvideo.mp4

Realsense D435i

detect_RS.py

import argparse
import time
from pathlib import Path

import cv2
import torch
import torch.backends.cudnn as cudnn
from numpy import random

from models.experimental import attempt_load
from utils.datasets import LoadStreams, LoadImages
from utils.general import check_img_size, check_requirements, check_imshow, non_max_suppression, apply_classifier, \
    scale_coords, xyxy2xywh, strip_optimizer, set_logging, increment_path
from utils.plots import plot_one_box
from utils.torch_utils import select_device, load_classifier, time_synchronized, TracedModel

import pyrealsense2 as rs
import numpy as np

def detect(save_img=False):
    source, weights, view_img, save_txt, imgsz, trace = opt.source, opt.weights, opt.view_img, opt.save_txt, opt.img_size, not opt.no_trace

    # Directories
    save_dir = Path(increment_path(Path(opt.project) / opt.name, exist_ok=opt.exist_ok))  # increment run
    (save_dir / 'labels' if save_txt else save_dir).mkdir(parents=True, exist_ok=True)  # make dir

    # Initialize
    set_logging()
    device = select_device(opt.device)
    half = device.type != 'cpu'  # half precision only supported on CUDA

    # Load model
    model = attempt_load(weights, map_location=device)  # load FP32 model
    stride = int(model.stride.max())  # model stride
    imgsz = check_img_size(imgsz, s=stride)  # check img_size

    if trace:
        model = TracedModel(model, device, opt.img_size)

    if half:
        model.half()  # to FP16

    # Second-stage classifier
    classify = False
    if classify:
        modelc = load_classifier(name='resnet101', n=2)  # initialize
        modelc.load_state_dict(torch.load('weights/resnet101.pt', map_location=device)['model']).to(device).eval()

    # Get names and colors
    names = model.module.names if hasattr(model, 'module') else model.names
    colors = [[random.randint(0, 255) for _ in range(3)] for _ in names]

    # Run inference
    if device.type != 'cpu':
        model(torch.zeros(1, 3, imgsz, imgsz).to(device).type_as(next(model.parameters())))  # run once
    old_img_w = old_img_h = imgsz
    old_img_b = 1

    config = rs.config()
    config.enable_stream(rs.stream.color, 640, 480, rs.format.bgr8, 30)
    config.enable_stream(rs.stream.depth, 640, 480, rs.format.z16, 30)

    pipeline = rs.pipeline()
    profile = pipeline.start(config)

    align_to = rs.stream.color
    align = rs.align(align_to)

    while(True):
        #t0 = time.time()
        frames = pipeline.wait_for_frames()

        aligned_frames = align.process(frames)
        color_frame = aligned_frames.get_color_frame()
        depth_frame = aligned_frames.get_depth_frame()
        if not depth_frame or not color_frame:
            continue

        img = np.asanyarray(color_frame.get_data())
        depth_image = np.asanyarray(depth_frame.get_data())
        depth_colormap = cv2.applyColorMap(cv2.convertScaleAbs(depth_image, alpha=0.08), cv2.COLORMAP_JET)
        
        # Letterbox
        im0 = img.copy()
        img = img[np.newaxis, :, :, :]        

        # Stack
        img = np.stack(img, 0)

        # Convert
        img = img[..., ::-1].transpose((0, 3, 1, 2))  # BGR to RGB, BHWC to BCHW
        img = np.ascontiguousarray(img)


        img = torch.from_numpy(img).to(device)
        img = img.half() if half else img.float()  # uint8 to fp16/32
        img /= 255.0  # 0 - 255 to 0.0 - 1.0
        if img.ndimension() == 3:
            img = img.unsqueeze(0)

        # Warmup
        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]):
            old_img_b = img.shape[0]
            old_img_h = img.shape[2]
            old_img_w = img.shape[3]
            for i in range(3):
                model(img, augment=opt.augment)[0]

        # Inference
        t1 = time_synchronized()
        with torch.no_grad():   # Calculating gradients would cause a GPU memory leak
            pred = model(img, augment=opt.augment)[0]
        t2 = time_synchronized()

        # Apply NMS
        pred = non_max_suppression(pred, opt.conf_thres, opt.iou_thres, classes=opt.classes, agnostic=opt.agnostic_nms)
        t3 = time_synchronized()

        # Process detections
        for i, det in enumerate(pred):  # detections per image

            gn = torch.tensor(im0.shape)[[1, 0, 1, 0]]  # normalization gain whwh
            if len(det):
                # Rescale boxes from img_size to im0 size
                det[:, :4] = scale_coords(img.shape[2:], det[:, :4], im0.shape).round()

                # Print results
                for c in det[:, -1].unique():
                    n = (det[:, -1] == c).sum()  # detections per class
                    #s += f"{n} {names[int(c)]}{'s' * (n > 1)}, "  # add to string

                # Write results
                for *xyxy, conf, cls in reversed(det):
                    c = int(cls)  # integer class
                    label = f'{names[c]} {conf:.2f}'
                    plot_one_box(xyxy, im0, label=label, color=colors[int(cls)], line_thickness=2)
                    plot_one_box(xyxy, depth_colormap, label=label, color=colors[int(cls)], line_thickness=2)             

            # Print time (inference + NMS)
            #print(f'{s}Done. ({(1E3 * (t2 - t1)):.1f}ms) Inference, ({(1E3 * (t3 - t2)):.1f}ms) NMS')

            # Stream results
            cv2.imshow("Recognition result", im0)
            cv2.imshow("Recognition result depth",depth_colormap)
            if cv2.waitKey(1) & 0xFF == ord('q'):
                break


if __name__ == '__main__':
    parser = argparse.ArgumentParser()
    parser.add_argument('--weights', nargs='+', type=str, default='yolov7-tiny.pt', help='model.pt path(s)')
    parser.add_argument('--source', type=str, default='inference/images', help='source')  # file/folder, 0 for webcam
    parser.add_argument('--img-size', type=int, default=640, help='inference size (pixels)')
    parser.add_argument('--conf-thres', type=float, default=0.25, help='object confidence threshold')
    parser.add_argument('--iou-thres', type=float, default=0.45, help='IOU threshold for NMS')
    parser.add_argument('--device', default='', help='cuda device, i.e. 0 or 0,1,2,3 or cpu')
    parser.add_argument('--view-img', action='store_true', help='display results')
    parser.add_argument('--save-txt', action='store_true', help='save results to *.txt')
    parser.add_argument('--save-conf', action='store_true', help='save confidences in --save-txt labels')
    parser.add_argument('--nosave', action='store_true', help='do not save images/videos')
    parser.add_argument('--classes', nargs='+', type=int, help='filter by class: --class 0, or --class 0 2 3')
    parser.add_argument('--agnostic-nms', action='store_true', help='class-agnostic NMS')
    parser.add_argument('--augment', action='store_true', help='augmented inference')
    parser.add_argument('--update', action='store_true', help='update all models')
    parser.add_argument('--project', default='runs/detect', help='save results to project/name')
    parser.add_argument('--name', default='exp', help='save results to project/name')
    parser.add_argument('--exist-ok', action='store_true', help='existing project/name ok, do not increment')
    parser.add_argument('--no-trace', action='store_true', help='don`t trace model')
    opt = parser.parse_args()
    print(opt)
    #check_requirements(exclude=('pycocotools', 'thop'))

    with torch.no_grad():
        if opt.update:  # update all models (to fix SourceChangeWarning)
            for opt.weights in ['yolov7.pt']:
                detect()
                strip_optimizer(opt.weights)
        else:
            detect()

$ pip install pyrealsense
$ python3 detect_RS.py
# 사람만 검출하고 싶다면
$ python3 detect_RS.py --classes 0
# 지정한 class에 대한 object만 검출 가능

Training with MS COCO Dataset

Data Preparation

# yolov7 경로에서 실행
$ bash scripts/get_coco.sh

Single GPU Training

# train p5 models
python train.py --workers 8 --device 0 --batch-size 32 --data data/coco.yaml --img 640 640 --cfg cfg/training/yolov7.yaml --weights '' --name yolov7 --hyp data/hyp.scratch.p5.yaml

# train p6 models
python train_aux.py --workers 8 --device 0 --batch-size 16 --data data/coco.yaml --img 1280 1280 --cfg cfg/training/yolov7-w6.yaml --weights '' --name yolov7-w6 --hyp data/hyp.scratch.p6.yaml

만약 학습이 도중에 끊겼을 때 끊긴 epoch부터 다시 학습시키고 싶다면 --resume 인자를 붙여주면 runs/train/yolov7(가장 최근에 저장된 학습 폴더)/weights/last.pt 로부터 끊기기 전에 저장된 최근의 weight부터 학습을 시작할 수 있다. 또한 --resume 뒤에 특정 weight 파일 경로를 입력해주면 특정 epoch부터 학습을 재개할 수 있다.

Test

Github에 나와있는 성능과 MSCOCO dataset으로 직접 학습한 모델의 성능과의 비교 작업을 위해 학습을 시킨 후 학습한 weight로 test 진행

python test.py --data data/coco.yaml --img 640 --batch 32 --conf 0.001 --iou 0.65 --device 0 --weights runs/train/yolov71/weights/best.pt --name yolov7_640_val

왼쪽이 github에 나와있는 성능표이고, 오른쪽이 직접 학습한 모델을 평가한 성능표이다. 결과적으로 거의 유사한 성능이 도출되었다는 것을 알 수 있다.

Error message

Traceback (most recent call last):
  File "train.py", line 616, in <module>
    train(hyp, opt, device, tb_writer)
  File "train.py", line 245, in train
    dataloader, dataset = create_dataloader(train_path, imgsz, batch_size, gs, opt,
  File "/home/***/***/yolov7/utils/datasets.py", line 69, in create_dataloader
    dataset = LoadImagesAndLabels(path, imgsz, batch_size,
  File "/home/***/***/yolov7/utils/datasets.py", line 396, in __init__
    cache, exists = self.cache_labels(cache_path, prefix), False  # cache
  File "/home/***/***/yolov7/utils/datasets.py", line 521, in cache_labels
    torch.save(x, path)  # save for next time
  File "/opt/conda/envs/yolov7/lib/python3.8/site-packages/torch/serialization.py", line 369, in save
    with _open_file_like(f, 'wb') as opened_file:
  File "/opt/conda/envs/yolov7/lib/python3.8/site-packages/torch/serialization.py", line 230, in _open_file_like
    return _open_file(name_or_buffer, mode)
  File "/opt/conda/envs/yolov7/lib/python3.8/site-packages/torch/serialization.py", line 211, in __init__
    super(_open_file, self).__init__(open(name, mode))
PermissionError: [Errno 13] Permission denied: 'coco/train2017.cache'

Solution

coco 폴더에 root 권한을 주면 해결됨

$ sudo chmod 777 coco

---

[References]

yolov7 paper
yolov7 github
robot mania youtube