CondLaneNet: a Top-to-down Lane Detection Framework Based on Conditional Convolution

Benchmark Model Rank Results
lane-detection-on-culaneCondLaneNet-L(ResNet-101)#18F1 score: 79.48
lane-detection-on-culaneCondLaneNet-M(ResNet-34)#22F1 score: 78.74
lane-detection-on-culaneCondLaneNet-S(ResNet-18)#23F1 score: 78.14
lane-detection-on-curvelanesCondLaneNet-L(ResNet-101)#5F1 score: 86.10GFLOPs: 44.9Precision: 88.98Recall: 83.41FPS: 48
lane-detection-on-curvelanesCondLaneNet-M(ResNet-34)#7F1 score: 85.92GFLOPs: 19.7Precision: 88.29Recall: 83.68
lane-detection-on-curvelanesCondLaneNet-S(ResNet-18)#8F1 score: 85.09GFLOPs: 10.3Precision: 87.75Recall: 82.58
lane-detection-on-tusimpleCondLaneNet-L(ResNet-101)#7Accuracy: 96.54%F1 score: 97.24
lane-detection-on-tusimpleCondLaneNet(ResNet-18)#21Accuracy: 95.48%
lane-detection-on-tusimpleCondLaneNet-M(ResNet-34)#23Accuracy: 95.37%F1 score: 96.98
lane-detection-on-tusimpleCondLaneNet(ResNet-34)#30F1 score: 97.01