Body condition estimation on Sow/Gilt Body Condition Dataset (test)
1.3MAE Fat (mm)Human Ultrasound Std
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Human Ultrasound Std2026.06 | 1.3 | 2.02 | 2.29 | 1.87 | — | — | — | — | |
| PigFormer w. Pruned MaskDINOInference Stage 1 (ms/frame)=52.73, Inference Stage 2 (ms/frame)=0.50, Stage 1 Variant=Pruned MaskDINO (R18-50q-5L), Stage 1 Precision=fp16, Stage 2 Precision=fp322026.06 | 2.34 | 5.27 | 4.2 | 3.94 | 14.1 | 9.8 | 5.9 | 8.4 | |
| PigFormer w. UNetInference Stage 1 (ms/frame)=6.58, Inference Stage 2 (ms/frame)=0.50, Stage 1 Variant=UNet (MobileNetV3-Small encoder), Stage 1 Precision=fp32, Stage 2 Precision=fp322026.06 | 2.4 | 5.2 | 4.26 | 3.95 | 14.4 | 9.7 | 6 | 8.4 | |
| PigFormer w. MaskDINOInference Stage 1 (ms/frame)=106.92, Inference Stage 2 (ms/frame)=0.50, Stage 1 Variant=MaskDINO (R50-300q-9L), Stage 1 Precision=fp16, Stage 2 Precision=fp322026.06 | 2.43 | 5.01 | 4.19 | 3.87 | 14.6 | 9.3 | 5.9 | 8.2 | |
| ResNet-18 (single-stage)Inference Stage 2 (ms/frame)=2.88, Backbone=ResNet-18, Precision=fp322026.06 | 2.88 | 6.1 | 5.81 | 4.93 | 17.3 | 11.4 | 8.2 | 10.5 | |
| ViT-small (single-stage)Inference Stage 2 (ms/frame)=4.98, Backbone=ViT-small, Precision=fp322026.06 | 3.57 | 7.29 | 8.16 | 6.34 | 21.4 | 13.6 | 11.5 | 13.5 |