# Efficient Tire Wear and Defect Detection Algorithm Based on Deep Learning
Paper : https://www.koreascience.or.kr/article/JAKO202125761197586.pdf \
Project report : https://scienceon.kisti.re.kr/commons/util/originalView.do
<br/>
## environment
- cuda 10.2
- python 3.6.12
- torch 1.8.1
- torchvision 0.9.1
- tensorflow 1.14.0
- keras 2.1.0
```
pip install -r requirements.txt
```
<br/>
## detectDefect
This project is tire defect detection model using [@matterport Mask-RCNN balloon.py](https://github.com/matterport/Mask_RCNN.git). \
and start by reading this [blog post about the balloon color splash sample](https://engineering.matterport.com/splash-of-color-instance-segmentation-with-mask-r-cnn-and-tensorflow-7c761e238b46).
### 1. Create dataset
Generate data Annotation and save JSON file using VGG Image Annotator, [VIA](https://www.robots.ox.ac.uk/~vgg/software/via/).
Each mask is set of polygon points.
### 2. Download a pretrained model
You can download pre-trained COCO weights (mask_rcnn_coco.h5) from the [releases page](https://github.com/matterport/Mask_RCNN/releases).
### 3. Train
```
python train.py --command train
--weights coco
--dataset [path/to/dataset]
--image [path/to/image]
```
### 4. Visualize
Visualize bounding box and defect masks
```
python train.py --command splash --weights [path/to/checkpoint] --dataset [path/to/dataset] --image [path/to/image]
```
![defect image](defect.png "defect result")
<br/>
## detectWear
### 1. Create tread mask
Prepare images and Gaussian tread attention masks. The masks are generated by Mask-RCNN.
<br/>
![mask image](mask.png "attention mask")
### 2. Train
```
python train.py
```
### 3. Test
```
python test.py --resume [path/to/checkpoint]
```
### 4. Class Activation Map
```
python cam.py --dataset_path [path/to/dataset]
--dataset [path/to/dataset/folder]
--model_path [path/to/checkpoint/folder]
--model_name [path/to/checkpoint.pth]
```
![cam image](cam.png "CAM result")
没有合适的资源?快使用搜索试试~ 我知道了~
资源推荐
资源详情
资源评论
收起资源包目录
【毕业设计】缺陷检测-基于深度学习实现的高效轮胎磨损+缺陷检测算法实现python源码.zip (34个子文件)
code
detectWear
parse_config.py 6KB
cam.py 4KB
utils
__init__.py 21B
util.py 4KB
data_loaders.py 4KB
base
__init__.py 56B
base_model.py 673B
base_trainer.py 6KB
logger
__init__.py 51B
visualization.py 3KB
logger_config.json 900B
logger.py 772B
trainer
__init__.py 24B
trainer.py 5KB
saved
cam
cam_example.png 59KB
model
loss.py 328B
model.py 16KB
metric.py 580B
requirements.txt 175B
config.json 2KB
train.py 3KB
test.py 3KB
mask.png 13KB
detectDefect
train_defect.py 15KB
mrcnn
utils.py 34KB
__init__.py 2B
model.py 127KB
parallel_model.py 7KB
visualize.py 19KB
config.py 9KB
requirements.txt 175B
cam.png 61KB
defect.png 6.04MB
README.md 2KB
共 34 条
- 1
资源评论
熬夜写代码的平头哥
- 粉丝: 3418
- 资源: 7310
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功