# self_driving_algorithms_deployment

**Repository Path**: Bin_lj/self_driving_algorithms_deployment

## Basic Information

- **Project Name**: self_driving_algorithms_deployment
- **Description**: 对自动驾驶相关的深度学习算法项目重构，仅用于推理和部署，不用于训练自己的数据集
- **Primary Language**: Unknown
- **License**: MIT
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2023-09-18
- **Last Updated**: 2023-09-21

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# tutorial
## lane detect
### start
if you want run ultra fast lane detection, please clone this hub in your computer.
1. mkdir 'weigths' in the dir of Ultra_Fast_lane_Detection download the weight file 'culane_18.pth'
2. move culane_18.pth into 'weights'
3. download minidata of CULane (***driver_37_30frame.tar.gz and laneseg_label_w16_test.zip are necessary***). 
4. extract minidate and move in the 'culane' directory,the The directory structure is as follows.  
   ![dir structure](lane_detection/Ultra_Fast_Lane_Detection_dir_structure.png)
### run
1. you can run main.py to detect lane,the result will be saved in 'output'.
2. you can run main.py to cheakout the detect result.
### show
the CUlane label as show below。
![gt_sample](lane_detection/culane_gt.png)
the predect result sample as show below:
![result_samppe](lane_detection/Ultra_Fast_Lane_Detection_Culane.png)
### requestments
```
numpy,scipy,torch,python-opencv,torchvision,tqdm
```

### download
the downland link is (https://www.123pan.com/s/evBjVv-b8g0.html)

## panoptic driving perception

### start
if you want run ultra fast lane detection, please clone this hub in your computer.    

### run
1. run main.py, you can get some result imgs in the dir 'runs'
2. run demo.py, it will save all result in the dir 'inference'
**Notes:** all files are uploaded,you just need run the script.  
***if you want to run all***,please download [all val data](https://www.123pan.com/s/evBjVv-B0g0.html)
,and move in the dir 'data_dir'.

### show
the test video result as show below:
[!show_mvie](https://gitee.com/Bin_lj/self_driving_algorithms_deployment/blob/master/panoptic-driving-perception/show.mp4)

### requestments
```
numpy, scipy, torch, python-opencv, torchvision, tqdm, yacs
```