Semantic Segmentation on IDD original size (val)
78.9mIoUDeeplabv3+
Evaluation Results
| Method | Links | |
|---|---|---|
| Deeplabv3+Resolution=1080 x 1920, Inference scale=single-scale, #FPS=3.5, #Params=40.5M2022.07 | 78.9 | |
| OCRNetResolution=1080 x 1920, Inference scale=single-scale, #FPS=4.2, #Params=39.0M2022.07 | 78.1 | |
| PSPNetResolution=1080 x 1920, Inference scale=single-scale, #FPS=5.2, #Params=31.1M2022.07 | 77.6 | |
| EMANetResolution=1080 x 1920, Inference scale=single-scale, #FPS=4.4, #Params=34.8M2022.07 | 77.2 | |
| SFNetBackbone=ResNet-18, Resolution=1080 x 1920, Inference scale=single-scale, #FPS=20.2, #Params=12.87M2022.07 | 76.8 | |
| SFNet-LiteBackbone=STDC-2, Resolution=1080 x 1920, Inference scale=single-scale, #FPS=26.2, #Params=13.7M2022.07 | 76.8 | |
| DANetResolution=1080 x 1920, Inference scale=single-scale, #FPS=3.2, #Params=48.1M2022.07 | 76.6 | |
| STDC2Resolution=1080 x 1920, Inference scale=single-scale, #FPS=24.8, #Params=16.1M2022.07 | 76.3 | |
| SFNetBackbone=DF2, Resolution=1080 x 1920, Inference scale=single-scale, #FPS=37.4, #Params=19.63M2022.07 | 76.3 | |
| SFNet-LiteBackbone=ResNet-18, Resolution=1080 x 1920, Inference scale=single-scale, #FPS=26.8, #Params=12.3M2022.07 | 76.2 | |
| SFNetBackbone=DF1, Resolution=1080 x 1920, Inference scale=single-scale, #FPS=65.8, #Params=9.03M2022.07 | 75.8 | |
| STDC1Resolution=1080 x 1920, Inference scale=single-scale, #FPS=30.8, #Params=12.0M2022.07 | 75.5 | |
| BiSeNet-V1Backbone=ResNet-18, Resolution=1080 x 1920, Inference scale=single-scale, #FPS=24.5, #Params=12.9M2022.07 | 74.4 | |
| Bi-AlignResolution=1080 x 1920, Inference scale=single-scale, #FPS=30.2, #Params=19.2M2022.07 | 73.9 | |
| ICNetResolution=1080 x 1920, Inference scale=single-scale, #FPS=37.5, #Params=26.5M2022.07 | 73.8 | |
| DF2-SegResolution=1080 x 1920, Inference scale=single-scale, #FPS=50.8, #Params=18.88M2022.07 | 67.9 | |
| DF1-SegResolution=1080 x 1920, Inference scale=single-scale, #FPS=79.2, #Params=8.55M2022.07 | 63.4 |