一、算法笔记

二、代码复现

2.1 搭建环境

创建虚拟环境

1
2
conda create -n yolov8 python=3.7
conda activate yolov8

安装PyTorch1.8.0

1
conda install pytorch==1.8.0 torchvision==0.9.0 torchaudio==0.8.0 cudatoolkit=11.1 -c pytorch -c conda-forge

下载作者开源的程序，并安装其他依赖

1
2
3
git clone https://github.com/ultralytics/ultralytics
cd ultralytics
pip install -r requirements.txt

2.2 命令行使用教程

（1）语法规则

1
2
3
4
5
yolo TASK MODE ARGS

Where   TASK (optional) is one of [detect, segment, classify]
        MODE (required) is one of [train, val, predict, export, track]
        ARGS (optional) are any number of custom 'arg=value' pairs like 'imgsz=320' that override defaults.

（2）训练

1
yolo train data=coco128.yaml model=yolov8n.pt epochs=10 lr0=0.01

（3）预测

1
yolo predict model=yolov8n-seg.pt source='https://youtu.be/Zgi9g1ksQHc' imgsz=320

（4）评价

1
yolo val model=yolov8n.pt data=coco128.yaml batch=1 imgsz=640

2.3 Python使用教程

（1）训练

1
2
3
4
from ultralytics import YOLO

model = YOLO('yolov8n.pt') # 从预训练模型开始
model.train(epochs=5)

（2）评价

1
2
3
4
5
from ultralytics import YOLO

model = YOLO("model.pt")
model.val()  # 使用model.pt的data yaml进行评价
model.val(data='coco128.yaml')  # 或指定数据进行评价

（3）预测

获取预测结果

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
from ultralytics import YOLO
from PIL import Image
import cv2

model = YOLO("model.pt")
# 接受所有类型 - image/dir/Path/URL/video/PIL/ndarray. 0 for webcam
# 从摄像头
results = model.predict(source="0")
# 从文件夹
results = model.predict(source="folder", show=True) # Display preds. Accepts all YOLO predict arguments

# 从PIL图像
im1 = Image.open("bus.jpg")
results = model.predict(source=im1, save=True)  # save plotted images

# 从ndarray
im2 = cv2.imread("bus.jpg")
results = model.predict(source=im2, save=True, save_txt=True)  # save predictions as labels

# 从PIL/ndarray的列表
results = model.predict(source=[im1, im2])

预测结果分析（results会包含预测所有结果的列表，当有很多图像的时候要注意避免内存溢出，特别是在实例分割时）

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
# 1. return as a list
results = model.predict(source="folder")

# 2.  return as a generator (stream=True)
results = model.predict(source=0, stream=True)

for result in results:
    # Detection
    result.boxes.xyxy   # box with xyxy format, (N, 4)
    result.boxes.xywh   # box with xywh format, (N, 4)
    result.boxes.xyxyn  # box with xyxy format but normalized, (N, 4)
    result.boxes.xywhn  # box with xywh format but normalized, (N, 4)
    result.boxes.conf   # confidence score, (N, 1)
    result.boxes.cls    # cls, (N, 1)

    # Segmentation
    result.masks.data      # masks, (N, H, W)
    result.masks.xy        # x,y segments (pixels), List[segment] * N
    result.masks.xyn       # x,y segments (normalized), List[segment] * N

    # Classification
    result.probs     # cls prob, (num_class, )

# Each result is composed of torch.Tensor by default, 
# in which you can easily use following functionality:
result = result.cuda()
result = result.cpu()
result = result.to("cpu")
result = result.numpy()

2.3 数据集制作（实例分割）

（1）使用Labelme创建实例分割数据集

安装labelme

1
pip install labelme

安装完成后直接在命令行输入labelme即可打开。

使用label进行标注，将生成的json文件和原始图像jpg，放入同一个文件夹中。

（2）Labelme格式转COCO格式

参考pypi的labelme2coco包

1
2
3
pip install labelme2coco
# 或者使用清华源
pip install -i https://pypi.tuna.tsinghua.edu.cn/simple labelme2coco
1
labelme2coco path/to/labelme/dir --train_split_rate 0.85

（3）COCO格式转YOLO格式

使用labelme制作实例分割的coco格式数据集，然后使用ultralytics/JSON2YOLO项目将json文件转换成yolo的训练格式。

1
2
git clone https://github.com/ultralytics/JSON2YOLO.git
cd JSON2YOLO

创建一个虚拟环境，然后使用下面的命令安装依赖

1
pip install -r requirements.txt

修改general_json2yolo.py的第387行，设置为刚才得到的COCO注释的位置，例如：

1
2
3
4
if source == 'COCO':
	convert_coco_json('datasets/20230223_Phone_4Obj_Coco/annotations',  # directory with *.json
					  use_segments=True,
					  cls91to80=True)

修改general_json2yolo.py的第289行，因为不是coco80中的物体类型，是自己设置的，因此需要修改：

1
2
# cls = coco80[ann['category_id'] - 1] if cls91to80 else ann['category_id'] - 1  # class
cls = ann['category_id']  # class

运行程序将COCO格式json文件转换为YOLO格式txt。

1
python general_json2yolo.py

结果保存在new_dir中，需要手动把images复制过去。最后得到的数据集如下：

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
data_root
├── images
	├── train2017
		├── youimagename.jpg
		└── ...
    └── val2017
		├── youimagename.jpg
		└── ...
└── labels
	├── train2017
		├── youimagename.txt
		└── ...
    └── val2017
		├── youimagename.txt
		└── ...

（4）创建数据集的YAML文件

打开目录ultralytics/datasets，复制一份其中的coco128-seg.yaml，重命名为custom-seg.yaml，然后根据自己的数据集进行修改。

例如：

1
2
3
4
5
6
7
8
9
10
11
12
# 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: /media/mahaofei/OneTouch/Dataset/Program_data/image_processing/ultralytics/20230223_Phone_4Obj_YOLO  # 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
names:
  0: ammeter
  1: coffeebox
  2: realsensebox
  3: sucker

2.4 开始训练

新建一个python文件如train.py，添加内容如下：

1
2
3
4
5
6
7
8
9
from ultralytics import YOLO

# Load a model
# model = YOLO('yolov8n-seg.yaml')  # build a new model from YAML
model = YOLO('yolov8n-seg.pt')  # load a pretrained model (recommended for training)
# model = YOLO('yolov8n-seg.yaml').load('yolov8n.pt')  # build from YAML and transfer weights

# Train the model
model.train(data='custom-seg.yaml', epochs=100, imgsz=3904, batch=1)

2.5 结果预测