Scrap-grade classification on Scrap Metal railcar level (test)
79RecallSwin 2 B
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Swin 2 BPre-training=ImageNet-1K, Learning_Paradigm=Multi-task learning (MTL)2026.02 | 79 | 80 | 80 | 79 | |
| Swin 2Architecture=Swin Transformer, Approach=Multi-Instance Learning2026.02 | 68 | 73 | 70 | 73 | |
| ViT B 32Pre-training=ImageNet-1K2026.02 | 66 | 70 | 67 | 63 | |
| ViT B 16Pre-training=ImageNet-1K2026.02 | 65 | 68 | 68 | 64 | |
| ResNeXt 101Pre-training=ImageNet-1K2026.02 | 62 | 67 | 53 | 56 | |
| ViTArchitecture=Vision Transformer2026.02 | 61 | 63 | 64 | 64 | |
| EfficientNArchitecture=EfficientNet2026.02 | 60 | 61 | 63 | 61 | |
| ResNeXtArchitecture=ResNeXt2026.02 | 59 | 61 | 65 | 63 | |
| ResNetArchitecture=ResNet2026.02 | 53 | 56 | 57 | 55 | |
| EfficientNet 7 BPre-training=ImageNet-1K2026.02 | 52 | 54 | 51 | 52 | |
| ResNet 50Pre-training=ImageNet-1K2026.02 | 51 | 27 | 41 | 43 |