Yutao Song

CLRDFNet: enhancing Cross-Layer Refinement Network with deep feature fusion for lane detection

Extensive experiments demonstrate the superiority of CLRDFNet, which achieves a 96.90% Accuracy and 97.96% F1 on TuSimple with ResNet34 backbone, alongside 56.98% mF1 and 82.76% CULane with DLA34 backbone.