Image Classification on ImageNet 1K (val) (Efficiency & Accuracy Delta)
-50.06Delta Parameters (%)BALF-P-0.5
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| BALF-P-0.5Backbone=ResNet-50, Objective=Parameter count, Target=0.52025.09 | -50.06 | -28.17 | -7.79 | |
| BALF-F-0.5Backbone=ResNet-50, Objective=FLOPs, Target=0.52025.09 | -43.84 | -50.03 | -13.96 | |
| T-ALSBackbone=ResNet-50, Constraint=High Params reduction2025.09 | -33.33 | — | -20.14 | |
| T-ALS-SBackbone=ResNet-50, Constraint=High Params reduction2025.09 | -33.33 | — | -6.9 | |
| BALF-P-0.7Backbone=ResNet-50, Objective=Parameter count, Target=0.72025.09 | -29.95 | -13.6 | -1.97 | |
| T-ALSBackbone=ResNet-50, Constraint=Low Params reduction2025.09 | -27.54 | — | -4.49 | |
| T-ALS-SBackbone=ResNet-50, Constraint=Low Params reduction2025.09 | -27.54 | — | -2.57 | |
| BALF-P-0.8Backbone=ResNet-50, Objective=Parameter count, Target=0.82025.09 | -21.9 | -9.03 | -1.11 | |
| BALF-F-0.7Backbone=ResNet-50, Objective=FLOPs, Target=0.72025.09 | -21.66 | -30.01 | -3.5 | |
| IFMBackbone=ResNet-50, Constraint=High Params reduction2025.09 | -20.42 | — | -10.45 | |
| DFPCBackbone=ResNet-50, Constraint=High Params reduction2025.09 | -10.92 | — | -13.88 | |
| ITVSPBackbone=ResNet-50, Constraint=High Params reduction2025.09 | -9.98 | — | -10.21 | |
| BALF-F-0.8Backbone=ResNet-50, Objective=FLOPs, Target=0.82025.09 | -9.73 | -20.03 | -1.69 | |
| DFPCBackbone=ResNet-50, Constraint=Low Params reduction2025.09 | -5.66 | — | -5.78 | |
| IFMBackbone=ResNet-50, Constraint=Low Params reduction2025.09 | -5.65 | — | -1.46 | |
| ITVSPBackbone=ResNet-50, Constraint=Low Params reduction2025.09 | -4.76 | — | -3.02 |