SSUL: Semantic Segmentation with Unknown Label for Exemplar-based Class-Incremental Learning

Open paper
Benchmark Model Rank Results
disjoint-10-1-on-pascal-voc-2012SSUL-M#1mIoU: 53.50
disjoint-10-1-on-pascal-voc-2012SSUL#2mIoU: 50.87
disjoint-15-1-on-pascal-voc-2012SSUL-M#2mIoU: 68.58
disjoint-15-1-on-pascal-voc-2012SSUL#3mIoU: 64.01
disjoint-15-5-on-pascal-voc-2012SSUL-M#2Mean IoU: 69.83
disjoint-15-5-on-pascal-voc-2012SSUL#3Mean IoU: 69.10
overlapped-10-1-on-pascal-voc-2012SSUL-M#3mIoU: 64.12
overlapped-10-1-on-pascal-voc-2012SSUL#6mIoU: 59.25
overlapped-100-5-on-ade20kSSUL-M#2mIoU: 34.56
overlapped-100-5-on-ade20kSSUL#3mIoU: 32.48
overlapped-100-50-on-ade20kSSUL-M#4mIoU: 34.37
overlapped-100-50-on-ade20kSSUL#5mIoU: 33.58
overlapped-15-1-on-pascal-voc-2012SSUL-M#4mIoU: 71.37
overlapped-15-1-on-pascal-voc-2012SSUL#5mIoU: 67.61
overlapped-15-5-on-pascal-voc-2012SSUL-M#4Mean IoU (val): 73.02
overlapped-15-5-on-pascal-voc-2012SSUL#7Mean IoU (val): 71.22
overlapped-50-50-on-ade20kSSUL-M#5mIoU: 29.77
overlapped-50-50-on-ade20kSSUL#6mIoU: 29.56

Rank counts only results with a code link.