CLRNetV2: A Faster and Stronger Lane Detector

Benchmark Model Rank Results
lane-detection-on-culaneCLRNetV2 (DLA34)–F1 score: 80.68mF1: 57.27
lane-detection-on-culaneCLRNetV2 (ResNet101)–F1 score: 80.43
lane-detection-on-culaneCLRNetV2 (ResNet18)–F1 score: 79.68
lane-detection-on-culaneCLRNetV2 (ResNet18-lite)–F1 score: 78.66
lane-detection-on-culaneCLRNetV2 (ResNet34)–F1 score: 79.94
lane-detection-on-curvelanesCLRNetV2 (ResNet101)–F1 score: 87.81
lane-detection-on-tusimpleCLRNetV2 (ResNet18)–Accuracy: 96.99F1 score: 97.90
lane-detection-on-tusimpleCLRNetV2 (ResNet34)–Accuracy: 96.88F1 score: 97.95