CANet: Curved Guide Line Network with Adaptive Decoder for Lane Detection

Benchmark Model Rank Results
lane-detection-on-culaneCANet-L(ResNet101)–F1 score: 79.86
lane-detection-on-culaneCANet-M–F1 score: 79.16
lane-detection-on-culaneCANet-S(ResNet18)–F1 score: 78.46
lane-detection-on-curvelanesCANet-L–F1 score: 87.87Precision: 91.69
lane-detection-on-curvelanesCANet-L(ResNet101)–GFLOPs: 45.7Recall: 84.36
lane-detection-on-curvelanesCANet-M–F1 score: 87.19GFLOPs: 22.6Precision: 91.53Recall: 83.25
lane-detection-on-curvelanesCANet-S–F1 score: 86.57GFLOPs: 13.1Precision: 91.37Recall: 82.25
lane-detection-on-tusimpleCANet-L(ResNet101)–Accuracy: 96.76%F1 score: 97.77
lane-detection-on-tusimpleCANet-M–Accuracy: 96.66%F1 score: 97.44
lane-detection-on-tusimpleCANet-S–Accuracy: 96.56%F1 score: 97.51