ICNet for Real-Time Semantic Segmentation on High-Resolution Images

Benchmark Model Rank Results
dichotomous-image-segmentation-on-dis-te1ICNet#17max F-Measure: 0.631weighted F-measure: 0.535MAE: 0.095
dichotomous-image-segmentation-on-dis-te2ICNet#15max F-Measure: 0.716weighted F-measure: 0.627MAE: 0.095
dichotomous-image-segmentation-on-dis-te3ICNet#12max F-Measure: 0.752weighted F-measure: 0.664MAE: 0.091
dichotomous-image-segmentation-on-dis-te4ICNet#12max F-Measure: 0.749weighted F-measure: 0.663MAE: 0.099
dichotomous-image-segmentation-on-dis-vdICNet#14max F-Measure: 0.697weighted F-measure: 0.609MAE: 0.102
real-time-semantic-segmentation-on-camvidICNet#23mIoU: 67.1%Frame (fps): 27.8Time (ms): 36
real-time-semantic-segmentation-on-cityscapesICNet#29mIoU: 70.6%Frame (fps): 30.3Time (ms): 33
semantic-segmentation-on-bdd100k-valICNet#19mIoU: 52.4(39.5fps)
semantic-segmentation-on-cityscapesICNet#71Mean IoU (class): 70.6%
semantic-segmentation-on-trans10kICNet#14mIoU: 23.39%GFLOPs: 10.64