# YOLO调试记录


## data文件夹 yaml文件配置

该文件夹中配置的是数据集的路径及类别信息


```python
path: ../datasets/dimo4
train: D:\yolov5-master\datasets\dimo4\images\train
val: D:\yolov5-master\datasets\dimo4\images\val
test: # test images (optional)
# Classes
names:
  0: dimo
```

## 模型导出

### 命令行形式

```bash
python export.py --weights runs/train/exp4/weights/best.pt --img 640 --batch 1 --device 0 --include onnx --opset 15       
```


通过export.py导出的文件格式为onnx

weights地址: runs/train/exp4/weights/xxx.pt    best.pt为train.py训练后所得模型

在detect.py中batch为1，batch在detect中不能更改,所以在导出onnx模型文件时要设置batch 1     batch在detect中不能更改

![image-20241113133535978](./README.assets/image-20241113133535978.png)

![1](./README.assets/1.png)

### PyCharm配置

##### export.py

```bash
--weights
runs/train/exp4/weights/best.pt
--img
640
--device0
--include
onnx
--opset
15
```


![image-20241113182234260](C:\Users\91492\AppData\Roaming\Typora\typora-user-images\image-20241113182234260.png)

## 模型加载

### 命令行形式

```bash
python detect.py --weights runs\train\exp4\weights\best.onnx --data D:\yolov5- master\data\dimo4.yaml
```

![image-20241113133825646](./README.assets/image-20241113133825646.png)

![image-20241113133840896](./README.assets/image-20241113133840896.png)

![image-20241113133850391](./README.assets/image-20241113133850391.png)

### PyCharm配置

```bash
--weights
runs\train\exp4\weights\best.onnx
--data
D:\yolov5-master\data\dimo4.yaml
```
![image-20241113182433935](C:\Users\91492\AppData\Roaming\Typora\typora-user-images\image-20241113182433935.png)

在detect.py中

输入的模型为在train.py中训练好的模型   ：地址: runs/train/exp4/weights/xxx.pt      

或者 ：runs/train/exp4/weights/xxx.onnx


![image-20241113133825646](./README.assets/image-20241113133825646.png)

![image-20241113133840896](./README.assets/image-20241113133840896.png)

![image-20241113133850391](./README.assets/image-20241113133850391.png)