TwinLiteNetPlus: A Stronger Model for Real-time Drivable Area and Lane Segmentation

Benchmark Model Rank Results
drivable-area-detection-on-bdd100k-valTwinLiteNetPlus-Large#2mIoU: 92.9Params (M): 1.94
drivable-area-detection-on-bdd100k-valTwinLiteNetPlus-Medium#4mIoU: 92.0Params (M): 0.48
drivable-area-detection-on-bdd100k-valTwinLiteNetPlus-Small#8mIoU: 90.6Params (M): 0.12
drivable-area-detection-on-bdd100k-valTwinLiteNetPlus-Nano#10mIoU: 87.3Params (M): 0.03
lane-detection-on-bdd100k-valTwinLiteNetPlus-Large#1IoU (%): 34.2Accuracy (%): 81.9Params (M): 1.94
lane-detection-on-bdd100k-valTwinLiteNetPlus-Medium#2IoU (%): 32.3Accuracy (%): 79.1Params (M): 0.48
lane-detection-on-bdd100k-valTwinLiteNetPlus-Small#6IoU (%): 29.3Accuracy (%): 75.8Params (M): 0.12
lane-detection-on-bdd100k-valTwinLiteNetPlus-Nano#10IoU (%): 23.3Accuracy (%): 70.2Params (M): 0.03