Scene Classification on NYUv2 (test)
78.8bAccEMSAFormer
Evaluation results
| Method | Links | |
|---|---|---|
| EMSAFormerBackbone=SwinV2-T-128-Multi-Aug, Task(s)=Sem + Sce + Ins + Or2023.06 | 78.8 | |
| EMSAFormer (SwinV2-T-128-Multi-Aug single-task)Backbone=SwinV2-T-128-Multi-Aug, Task(s)=Scene Classification (Sce), Throughput (50W)=58.9 FPS, Throughput (30W)=40.7 FPS2023.06 | 78.66 | |
| EMSAFormerBackbone=SwinV2-T-128-Multi-Aug, Task(s)=Sem(SegFormer) + Sce + Ins + Or2023.06 | 77.7 | |
| EMSANetBackbone=2x ResNet101, Task(s)=Sem + Sce + Ins + Or2023.06 | 77.41 | |
| EMSANetBackbone=2x ResNet34-NBt1D, Task(s)=Sem + Sce + Ins + Or2023.06 | 76.46 |