In Defense of Pre-trained ImageNet Architectures for Real-time Semantic Segmentation of Road-driving Images

Benchmark Model Rank Results
real-time-semantic-segmentation-on-cityscapesSwiftNetRN-18#12mIoU: 75.5%Frame (fps): 39.9
semantic-segmentation-on-cityscapesSwiftNetRN-18#60Mean IoU (class): 75.5%
semantic-segmentation-on-dada-segSwiftNet (ResNet-18)#19mIoU: 20.5
semantic-segmentation-on-densepassSwiftNet (Cityscapes)#33mIoU: 25.67%
semantic-segmentation-on-eventscapeSwiftNet#12mIoU: 36.67
semantic-segmentation-on-zju-rgb-pSwiftNet (RGB)#11mIoU: 80.3