Semantic Segmentation on Stanford2D3D fold-1 (Area 5a and 5b)
45.73mIoUTrans4Trans-M
Evaluation Results
| Method | Links | |
|---|---|---|
| Trans4Trans-MEncoder=PVT-M [63], Decoder=TPM / Single-head, GFLOPs=34.38, MParams=43.65, Input Size=512x5122021.07 | 45.73 | |
| Trans4Trans-MEncoder=PVT-M [63], Decoder=TPM / Dual-head, GFLOPs=35.17, MParams=44.04, Input Size=512x5122021.07 | 45.15 | |
| Trans4Trans-SEncoder=PVT-S [63], Decoder=TPM / Single-head, GFLOPs=19.92, MParams=23.95, Input Size=512x5122021.07 | 44.47 | |
| Trans2Seg-SEncoder=R50 [24], Decoder=Transformer [70], GFLOPs=40.98, MParams=30.53, Input Size=512x5122021.07 | 43.83 | |
| Trans4Trans-SEncoder=PVT-S [63], Decoder=TPM / Dual-head, GFLOPs=20.69, MParams=24.34, Input Size=512x5122021.07 | 43.45 | |
| Trans2Seg-TEncoder=R34 [24], Decoder=Transformer [70], GFLOPs=30.26, MParams=27.98, Input Size=512x5122021.07 | 42.91 | |
| PVT-MEncoder=PVT-M [63], Decoder=Transformer [70], GFLOPs=49, MParams=56.2, Input Size=512x5122021.07 | 42.49 | |
| Trans2Seg-TEncoder=R18 [24], Decoder=Transformer [70], GFLOPs=16.96, MParams=17.87, Input Size=512x5122021.07 | 42.07 | |
| PVT-SEncoder=PVT-S [63], Decoder=Transformer [70], GFLOPs=19.58, MParams=24.36, Input Size=512x5122021.07 | 41.89 | |
| Trans4Trans-TEncoder=PVT-T [63], Decoder=TPM / Single-head, GFLOPs=10.45, MParams=12.71, Input Size=512x5122021.07 | 41.28 | |
| PVT-TEncoder=PVT-T [63], Decoder=Transformer [70], GFLOPs=10.16, MParams=13.11, Input Size=512x5122021.07 | 41 | |
| Trans4Trans-TEncoder=PVT-T [63], Decoder=TPM / Dual-head, GFLOPs=11.22, MParams=13.1, Input Size=512x5122021.07 | 40.44 |