Lane Detection

Final project codes for Computer vision in the spring semester at NYCU. This is a basic lane detection using TuSimple dataset.

109-2 CV Final

• Zhi-Yi Chin [email protected]
• Shao-Yu Weng [email protected]
• Bo-Yu Cheng [email protected]

Technologies

Prerequisite:

conda environment
1080Ti or 2080Ti GPU device
pytorch
torchvision
visdom
cudatoolkit=10.1
opencv
sklearn
ujson
matplotlib

Method Description

Deep Learning Method

2 Hourglass Network + aggregation of SGPN loss

Traditional Method

• Method 1: Edge detection + Hough transform
• Method 2: Perspective transform

Results

License