目录
0 前言
1 环境配置
1.1 python环境
1.2 官方github的样例
1.2.1 打印检测结果
1.2.2 展示检测结果
2 运用detect.py进行检测
2.1 网络摄像头
2.2 将检测过程可视化
3 运用train.py进行训练
3.1 第一次报错
3.2 换一条命令
3.3 对比上面两条命令的数据集
3.4 第一次报错解决一半
未完
电脑:RTX3070、cuda-11.0，系统ubuntu18.04
官网：https://github.com/ultralytics/yolov5
yolov5注释大神：https://github.com/SCAU-HuKai/yolov5-5.x-annotations
其CSDN：https://blog.csdn.net/qq_38253797/article/details/119043919
下载代码：
git clone https://github.com/ultralytics/yolov5进入到下载目录：
#创建python3.7的环境
conda create -n yolov5py37 python=3.7
#安装gpu版本的pytorch
#官网链接：https://pytorch.org/get-started/previous-versions/
pip install torch==1.7.1+cu110 torchvision==0.8.2+cu110 torchaudio==0.7.2 -f https://download.pytorch.org/whl/torch_stable.html
#上面这个命令安装的包：
torch==1.7.1+cu110
torchvision==0.8.2+cu110
torchaudio==0.7.2
#安装其他需要的包
pip install -r requirements.txt
#安装包的版本可能不固定，但都是满足要求的版本创建文件inference.py
import torch
# Model
model = torch.hub.load('ultralytics/yolov5', 'yolov5s') # or yolov5m, yolov5l, yolov5x, custom
# Images
img = 'https://ultralytics.com/images/zidane.jpg' # or file, Path, PIL, OpenCV, numpy, list
# Inference
results = model(img)
# Results
results.print() # or .show(), .save(), .crop(), .pandas(), etc.在1.1的python环境中执行，终端打印输出的有yolov5软件信息、计算机硬件信息、下载.pt模型文件、模型概况、图像、检测结果。
python inference.py(yolov5py37) meng@meng:~/deeplearning/yolov5$ python inference.py
Downloading: "https://github.com/ultralytics/yolov5/archive/master.zip" to /home/meng/.cache/torch/hub/master.zip
Downloading https://ultralytics.com/assets/Arial.ttf to /home/meng/.config/Ultralytics/Arial.ttf...
fatal: 不是一个 git 仓库（或者任何父目录）：.git
YOLOv5 🚀 2022-3-12 torch 1.7.1+cu110 CUDA:0 (NVIDIA GeForce RTX 3070, 7960MiB)
Downloading https://github.com/ultralytics/yolov5/releases/download/v6.1/yolov5s.pt to yolov5s.pt...
100%|█████████████████████████████████████| 14.1M/14.1M [00:07<00:00, 2.06MB/s]
Fusing layers...
Model Summary: 213 layers, 7225885 parameters, 0 gradients, 16.5 GFLOPs
Adding AutoShape...
image 1/1: 720x1280 2 persons, 2 ties
Speed: 7411.2ms pre-process, 8.4ms inference, 1.2ms NMS per image at shape (1, 3, 384, 640)
1.2.2 展示检测结果inference.py最后一行，将print改为show
可选择的命令有：
python detect.py --source 0 # webcam
img.jpg # image
vid.mp4 # video
path/ # directory
path/*.jpg # glob
'https://youtu.be/Zgi9g1ksQHc' # YouTube
'rtsp://example.com/media.mp4' # RTSP, RTMP, HTTP stream
执行上面的命令会自动从最新的yolov5官网下载模型文件，并且将检测结果存在runs/detect
给电脑插一个相机（usb相机即可），笔记本可能会自带摄像头
python detect.py --source 0下面第二行应该是可以调整的参数，检测效果来看，帧率挺高的。
(yolov5py37) meng@meng:~/deeplearning/yolov5$ python detect.py --source 0
detect: weights=yolov5s.pt, source=0, data=data/coco128.yaml, imgsz=[640, 640], conf_thres=0.25, iou_thres=0.45, max_det=1000, device=, view_img=False, save_txt=False, save_conf=False, save_crop=False, nosave=False, classes=None, agnostic_nms=False, augment=False, visualize=False, update=False, project=runs/detect, name=exp, exist_ok=False, line_thickness=3, hide_labels=False, hide_conf=False, half=False, dnn=False
YOLOv5 🚀 v6.1-28-gc6b4f84 torch 1.7.1+cu110 CUDA:0 (NVIDIA GeForce RTX 3070, 7960MiB)
Fusing layers...
Model Summary: 213 layers, 7225885 parameters, 0 gradients, 16.5 GFLOPs
1/1: 0... Success (inf frames 640x480 at 30.00 FPS)
0: 480x640 1 person, 1 cup, 2 chairs, 2 tvs, Done. (0.501s)
------------------------省略
python detect.py --visualize上面这条命令使用的数据是默认图片，终端将检测的步骤打印输出：
在runs/detect/expn下面将步骤保存下来，如下：
下图是stage0：stage0_Conv_features.png，其他png类推
关于.npy格式文件，如：stage0_Conv_features.npy。打开方式为：新建一个python文件如下，new.py；
import numpy as np
test=np.load("/home/meng/deeplearning/yolov5/runs/detect/exp6/bus/stage0_Conv_features.npy")
print(test)运行python new.py即可看到里面的矩阵数据，但数据挺多的：
python train.py --data coco.yaml --cfg yolov5n.yaml --weights '' --batch-size 128报错为：
Traceback (most recent call last):
File "train.py", line 643, in <module>
main(opt)
File "train.py", line 539, in main
train(opt.hyp, opt, device, callbacks)
File "train.py", line 227, in train
prefix=colorstr('train: '), shuffle=True)
File "/home/meng/deeplearning/yolov5/utils/datasets.py", line 109, in create_dataloader
prefix=prefix)
File "/home/meng/deeplearning/yolov5/utils/datasets.py", line 433, in __init__
assert nf > 0 or not augment, f'{prefix}No labels in {cache_path}. Can not train without labels. See {HELP_URL}'
AssertionError: train: No labels in /home/meng/deeplearning/datasets/coco/train2017.cache. Can not train without labels. See https://github.com/ultralytics/yolov5/wiki/Train-Custom-Data删除和yolov5在同一级目录的刚下载的数据集：
参考：Train Custom Data · ultralytics/yolov5 Wiki · GitHub
python train.py --img 640 --batch 16 --epochs 3 --data coco128.yaml --weights yolov5s.pt可以跑通，尽管也报：Dataset not found, missing paths: ['/home/meng/deeplearning/datasets/coco128/images/train2017']
meng@meng:~/deeplearning/yolov5$ python train.py --img 640 --batch 16 --epochs 3 --data coco128.yaml --weights yolov5s.pt
train: weights=yolov5s.pt, cfg=, data=coco128.yaml, hyp=data/hyps/hyp.scratch-low.yaml, epochs=3, batch_size=16, imgsz=640, rect=False, resume=False, nosave=False, noval=False, noautoanchor=False, evolve=None, bucket=, cache=None, image_weights=False, device=, multi_scale=False, single_cls=False, optimizer=SGD, sync_bn=False, workers=8, project=runs/train, name=exp, exist_ok=False, quad=False, cos_lr=False, label_smoothing=0.0, patience=100, freeze=[0], save_period=-1, local_rank=-1, entity=None, upload_dataset=False, bbox_interval=-1, artifact_alias=latest
github: skipping check (offline), for updates see https://github.com/ultralytics/yolov5
YOLOv5 🚀 v6.1-28-gc6b4f84 torch 1.7.1+cu110 CUDA:0 (NVIDIA GeForce RTX 3070, 7960MiB)
hyperparameters: lr0=0.01, lrf=0.01, momentum=0.937, weight_decay=0.0005, warmup_epochs=3.0, warmup_momentum=0.8, warmup_bias_lr=0.1, box=0.05, cls=0.5, cls_pw=1.0, obj=1.0, obj_pw=1.0, iou_t=0.2, anchor_t=4.0, fl_gamma=0.0, hsv_h=0.015, hsv_s=0.7, hsv_v=0.4, degrees=0.0, translate=0.1, scale=0.5, shear=0.0, perspective=0.0, flipud=0.0, fliplr=0.5, mosaic=1.0, mixup=0.0, copy_paste=0.0
Weights & Biases: run 'pip install wandb' to automatically track and visualize YOLOv5 🚀 runs (RECOMMENDED)
TensorBoard: Start with 'tensorboard --logdir runs/train', view at http://localhost:6006/
Dataset not found, missing paths: ['/home/meng/deeplearning/datasets/coco128/images/train2017']
Downloading https://ultralytics.com/assets/coco128.zip to coco128.zip...
100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6.66M/6.66M [00:02<00:00, 2.44MB/s]
Dataset autodownload success, saved to /home/meng/deeplearning/datasets
from n params module arguments
0 -1 1 3520 models.common.Conv [3, 32, 6, 2, 2]
1 -1 1 18560 models.common.Conv [32, 64, 3, 2]
2 -1 1 18816 models.common.C3 [64, 64, 1]
3 -1 1 73984 models.common.Conv [64, 128, 3, 2]
4 -1 2 115712 models.common.C3 [128, 128, 2]
5 -1 1 295424 models.common.Conv [128, 256, 3, 2]
6 -1 3 625152 models.common.C3 [256, 256, 3]
7 -1 1 1180672 models.common.Conv [256, 512, 3, 2]
8 -1 1 1182720 models.common.C3 [512, 512, 1]
9 -1 1 656896 models.common.SPPF [512, 512, 5]
10 -1 1 131584 models.common.Conv [512, 256, 1, 1]
11 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
12 [-1, 6] 1 0 models.common.Concat [1]
13 -1 1 361984 models.common.C3 [512, 256, 1, False]
14 -1 1 33024 models.common.Conv [256, 128, 1, 1]
15 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
16 [-1, 4] 1 0 models.common.Concat [1]
17 -1 1 90880 models.common.C3 [256, 128, 1, False]
18 -1 1 147712 models.common.Conv [128, 128, 3, 2]
19 [-1, 14] 1 0 models.common.Concat [1]
20 -1 1 296448 models.common.C3 [256, 256, 1, False]
21 -1 1 590336 models.common.Conv [256, 256, 3, 2]
22 [-1, 10] 1 0 models.common.Concat [1]
23 -1 1 1182720 models.common.C3 [512, 512, 1, False]
24 [17, 20, 23] 1 229245 models.yolo.Detect [80, [[10, 13, 16, 30, 33, 23], [30, 61, 62, 45, 59, 119], [116, 90, 156, 198, 373, 326]], [128, 256, 512]]
Model Summary: 270 layers, 7235389 parameters, 7235389 gradients, 16.5 GFLOPs
Transferred 349/349 items from yolov5s.pt
Scaled weight_decay = 0.0005
optimizer: SGD with parameter groups 57 weight (no decay), 60 weight, 60 bias
train: Scanning '/home/meng/deeplearning/datasets/coco128/labels/train2017' images and labels...128 found, 0 missing, 2 empty, 0 corrupt: 100%|███████████████████████████████████████████████████████████████████████████████████████████████| 128/128 [00:00<00:00, 11289.71it/s]
train: New cache created: /home/meng/deeplearning/datasets/coco128/labels/train2017.cache
val: Scanning '/home/meng/deeplearning/datasets/coco128/labels/train2017.cache' images and labels... 128 found, 0 missing, 2 empty, 0 corrupt: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████| 128/128 [00:00<?, ?it/s]
Plotting labels to runs/train/exp8/labels.jpg...
AutoAnchor: 4.27 anchors/target, 0.994 Best Possible Recall (BPR). Current anchors are a good fit to dataset ✅
Image sizes 640 train, 640 val
Using 8 dataloader workers
Logging results to runs/train/exp8
Starting training for 3 epochs...
Epoch gpu_mem box obj cls labels img_size
0/2 3.3G 0.04377 0.06153 0.01789 226 640: 100%|██████████| 8/8 [00:03<00:00, 2.39it/s]
Class Images Labels P R mAP@.5 mAP@.5:.95: 100%|██████████| 4/4 [00:00<00:00, 6.85it/s]
all 128 929 0.759 0.618 0.717 0.474
Epoch gpu_mem box obj cls labels img_size
1/2 4.27G 0.04464 0.0689 0.01817 207 640: 100%|██████████| 8/8 [00:01<00:00, 4.49it/s]
Class Images Labels P R mAP@.5 mAP@.5:.95: 100%|██████████| 4/4 [00:00<00:00, 6.95it/s]
all 128 929 0.694 0.678 0.732 0.487
Epoch gpu_mem box obj cls labels img_size
2/2 4.27G 0.04492 0.06209 0.01751 241 640: 100%|██████████| 8/8 [00:01<00:00, 4.52it/s]
Class Images Labels P R mAP@.5 mAP@.5:.95: 100%|██████████| 4/4 [00:00<00:00, 6.98it/s]
all 128 929 0.704 0.674 0.737 0.49
3 epochs completed in 0.003 hours.
Optimizer stripped from runs/train/exp8/weights/last.pt, 14.9MB
Optimizer stripped from runs/train/exp8/weights/best.pt, 14.9MB
Validating runs/train/exp8/weights/best.pt...
Fusing layers...
Model Summary: 213 layers, 7225885 parameters, 0 gradients, 16.5 GFLOPs
Class Images Labels P R mAP@.5 mAP@.5:.95: 100%|██████████| 4/4 [00:03<00:00, 1.08it/s]
all 128 929 0.699 0.679 0.738 0.491
person 128 254 0.828 0.736 0.811 0.52
bicycle 128 6 0.778 0.593 0.648 0.407
car 128 46 0.655 0.454 0.576 0.248
motorcycle 128 5 0.571 0.8 0.866 0.705
airplane 128 6 0.921 1 0.995 0.736
bus 128 7 0.626 0.714 0.738 0.626
train 128 3 0.613 0.667 0.806 0.571
truck 128 12 0.454 0.417 0.495 0.27
boat 128 6 0.794 0.333 0.464 0.173
traffic light 128 14 0.648 0.266 0.362 0.216
stop sign 128 2 0.751 1 0.995 0.796
bench 128 9 0.633 0.556 0.624 0.231
bird 128 16 0.899 1 0.995 0.634
cat 128 4 0.704 1 0.995 0.797
dog 128 9 0.782 0.667 0.851 0.567
horse 128 2 0.723 1 0.995 0.672
elephant 128 17 0.945 0.882 0.934 0.694
bear 128 1 0.635 1 0.995 0.895
zebra 128 4 0.848 1 0.995 0.947
giraffe 128 9 0.704 0.778 0.94 0.687
backpack 128 6 0.751 0.5 0.779 0.362
umbrella 128 18 0.873 0.765 0.899 0.513
handbag 128 19 0.599 0.238 0.335 0.142
tie 128 7 0.708 0.714 0.81 0.498
suitcase 128 4 0.726 1 0.995 0.563
frisbee 128 5 0.688 0.8 0.8 0.7
skis 128 1 0.598 1 0.995 0.398
snowboard 128 7 0.796 0.714 0.848 0.567
sports ball 128 6 0.613 0.667 0.603 0.309
kite 128 10 0.777 0.698 0.629 0.249
baseball bat 128 4 0.381 0.5 0.4 0.135
baseball glove 128 7 0.527 0.429 0.457 0.309
skateboard 128 5 1 0.571 0.69 0.476
tennis racket 128 7 0.438 0.448 0.534 0.291
bottle 128 18 0.695 0.635 0.6 0.281
wine glass 128 16 0.605 1 0.916 0.469
cup 128 36 0.795 0.753 0.845 0.542
fork 128 6 0.866 0.333 0.445 0.314
knife 128 16 0.731 0.688 0.656 0.367
spoon 128 22 0.695 0.545 0.645 0.35
bowl 128 28 0.834 0.719 0.741 0.505
banana 128 1 0.465 1 0.995 0.298
sandwich 128 2 1 0 0.606 0.535
orange 128 4 0.801 1 0.995 0.703
broccoli 128 11 0.461 0.455 0.476 0.35
carrot 128 24 0.633 0.625 0.736 0.473
hot dog 128 2 0.436 1 0.828 0.712
pizza 128 5 0.833 0.8 0.962 0.677
donut 128 14 0.667 1 0.96 0.82
cake 128 4 0.823 1 0.995 0.846
chair 128 35 0.485 0.629 0.587 0.297
couch 128 6 1 0.761 0.881 0.54
potted plant 128 14 0.712 0.786 0.858 0.471
bed 128 3 0.43 0.287 0.597 0.374
dining table 128 13 0.798 0.608 0.598 0.389
toilet 128 2 0.791 1 0.995 0.895
tv 128 2 0.575 1 0.995 0.796
laptop 128 3 0.965 0.333 0.665 0.399
mouse 128 2 1 0 0.0923 0.0462
remote 128 8 0.806 0.625 0.635 0.54
cell phone 128 8 0.474 0.375 0.365 0.214
microwave 128 3 0.749 1 0.995 0.7
oven 128 5 0.413 0.4 0.44 0.289
sink 128 6 0.377 0.333 0.34 0.217
refrigerator 128 5 0.593 0.8 0.808 0.55
book 128 29 0.462 0.356 0.321 0.158
clock 128 9 0.671 0.778 0.879 0.728
vase 128 2 0.408 1 0.995 0.895
scissors 128 1 1 0 0.332 0.0663
teddy bear 128 21 0.764 0.667 0.787 0.496
toothbrush 128 5 0.802 1 0.962 0.621
Results saved to runs/train/exp8
训练结果如下：
这是3.1指令对应的数据集文件：coco.yaml。
# YOLOv5 🚀 by Ultralytics, GPL-3.0 license
# COCO 2017 dataset http://cocodataset.org by Microsoft
# Example usage: python train.py --data coco.yaml
# parent
# ├── yolov5
# └── datasets
# └── coco ← downloads here
# Train/val/test sets as 1) dir: path/to/imgs, 2) file: path/to/imgs.txt, or 3) list: [path/to/imgs1, path/to/imgs2, ..]
path: ../datasets/coco # dataset root dir
train: train2017.txt # train images (relative to 'path') 118287 images
val: val2017.txt # val images (relative to 'path') 5000 images
test: test-dev2017.txt # 20288 of 40670 images, submit to https://competitions.codalab.org/competitions/20794
# Classes
nc: 80 # number of classes
names: ['person', 'bicycle', 'car', 'motorcycle', 'airplane', 'bus', 'train', 'truck', 'boat', 'traffic light',
'fire hydrant', 'stop sign', 'parking meter', 'bench', 'bird', 'cat', 'dog', 'horse', 'sheep', 'cow',
'elephant', 'bear', 'zebra', 'giraffe', 'backpack', 'umbrella', 'handbag', 'tie', 'suitcase', 'frisbee',
'skis', 'snowboard', 'sports ball', 'kite', 'baseball bat', 'baseball glove', 'skateboard', 'surfboard',
'tennis racket', 'bottle', 'wine glass', 'cup', 'fork', 'knife', 'spoon', 'bowl', 'banana', 'apple',
'sandwich', 'orange', 'broccoli', 'carrot', 'hot dog', 'pizza', 'donut', 'cake', 'chair', 'couch',
'potted plant', 'bed', 'dining table', 'toilet', 'tv', 'laptop', 'mouse', 'remote', 'keyboard', 'cell phone',
'microwave', 'oven', 'toaster', 'sink', 'refrigerator', 'book', 'clock', 'vase', 'scissors', 'teddy bear',
'hair drier', 'toothbrush'] # class names
# Download script/URL (optional)
download: |
from utils.general import download, Path
# Download labels
segments = False # segment or box labels
dir = Path(yaml['path']) # dataset root dir
url = 'https://github.com/ultralytics/yolov5/releases/download/v1.0/'
urls = [url + ('coco2017labels-segments.zip' if segments else 'coco2017labels.zip')] # labels
download(urls, dir=dir.parent)
# Download data
urls = ['http://images.cocodataset.org/zips/train2017.zip', # 19G, 118k images
'http://images.cocodataset.org/zips/val2017.zip', # 1G, 5k images
'http://images.cocodataset.org/zips/test2017.zip'] # 7G, 41k images (optional)
download(urls, dir=dir / 'images', threads=3) 最下面有可选择下载的数据集，看出来19G、1G、7G比较大。对应执行3.1命令后，没有成功下载好（后缀为.cache）
相对比3.2指令对应的数据集文件coco128.yaml；文件中下载的数据在最后一行，没有直接标大小，感觉也不是很大（后来程序运行成功发现确实不大）
# YOLOv5 🚀 by Ultralytics, GPL-3.0 license
# COCO128 dataset https://www.kaggle.com/ultralytics/coco128 (first 128 images from COCO train2017) by Ultralytics
# Example usage: python train.py --data coco128.yaml
# parent
# ├── yolov5
# └── datasets
# └── coco128 ← downloads here
# Train/val/test sets as 1) dir: path/to/imgs, 2) file: path/to/imgs.txt, or 3) list: [path/to/imgs1, path/to/imgs2, ..]
path: ../datasets/coco128 # dataset root dir
train: images/train2017 # train images (relative to 'path') 128 images
val: images/train2017 # val images (relative to 'path') 128 images
test: # test images (optional)
# Classes
nc: 80 # number of classes
names: ['person', 'bicycle', 'car', 'motorcycle', 'airplane', 'bus', 'train', 'truck', 'boat', 'traffic light',
'fire hydrant', 'stop sign', 'parking meter', 'bench', 'bird', 'cat', 'dog', 'horse', 'sheep', 'cow',
'elephant', 'bear', 'zebra', 'giraffe', 'backpack', 'umbrella', 'handbag', 'tie', 'suitcase', 'frisbee',
'skis', 'snowboard', 'sports ball', 'kite', 'baseball bat', 'baseball glove', 'skateboard', 'surfboard',
'tennis racket', 'bottle', 'wine glass', 'cup', 'fork', 'knife', 'spoon', 'bowl', 'banana', 'apple',
'sandwich', 'orange', 'broccoli', 'carrot', 'hot dog', 'pizza', 'donut', 'cake', 'chair', 'couch',
'potted plant', 'bed', 'dining table', 'toilet', 'tv', 'laptop', 'mouse', 'remote', 'keyboard', 'cell phone',
'microwave', 'oven', 'toaster', 'sink', 'refrigerator', 'book', 'clock', 'vase', 'scissors', 'teddy bear',
'hair drier', 'toothbrush'] # class names
# Download script/URL (optional)
download: https://ultralytics.com/assets/coco128.zip将3.3中coco.yaml文件里面的链接文件提前下载下来
http://images.cocodataset.org/zips/train2017.zip # 19G, 118k images
http://images.cocodataset.org/zips/val2017.zip # 1G, 5k images
http://images.cocodataset.org/zips/test2017.zip解压到这个文件夹下
同时将coco.yaml 中download部分删除掉（备份好）。
训练：
python train.py --data coco.yaml --cfg yolov5n.yaml --weights '' --batch-size 128这时可以正常找到图片文件，但是cuda out of memory了；我的显存只有8g
RuntimeError: CUDA out of memory. Tried to allocate 200.00 MiB (GPU 0; 7.77 GiB total capacity; 5.70 GiB already allocated; 177.62 MiB free; 5.92 GiB reserved in total by PyTorch)修改batch-size为16，可以运行，gpu_memory占用
修改为40，gpu_mem占用：
训练效果如下：（现在这样的参数--挺慢的）
