Semantic Segmentation Efficiency on 256x256 Input
9.08Params (M)OneNet
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| OneNetarchitecture=encoder and decoder replaced, downsampling layers=42024.11 | 9.08 | 36.3 | 22.92 | 799.01 | |
| MobileNet2024.11 | 14.4 | 57.47 | 83.96 | 671.09 | |
| OneNetarchitecture=encoder replaced, downsampling layers=42024.11 | 16.39 | 65.42 | 78.42 | 639.63 | |
| ResNetlayers=342024.11 | 25.05 | 98.07 | 29.4 | 241.17 | |
| U-Netdownsampling layers=42024.11 | 31.04 | 124.03 | 104.72 | 509.61 | |
| OneNetarchitecture=encoder and decoder replaced, downsampling layers=52024.11 | 36.38 | 145.47 | 39 | 885 | |
| OneNetarchitecture=encoder replaced, downsampling layers=52024.11 | 65.73 | 262.63 | 98.82 | 656.41 | |
| ResNetlayers=502024.11 | 74.07 | 287.83 | 84.98 | 450.36 | |
| U-Netdownsampling layers=52024.11 | 124.42 | 497.41 | 130.8 | 524.29 |