Image Classification on FryNet (test)
100AccuracyFryNet
Evaluation Results
| Method | Links | |
|---|---|---|
| FryNetBackbone=MiT-B2, Params=31.0M, GFLOPs=30.3, FPS=47.1, Modality=RGB + thermal2026.04 | 100 | |
| DINOv2Backbone=ViT-B, Params=87.8M, GFLOPs=119.0, FPS=46.4, Modality=thermal only2026.04 | 80.1 | |
| CMNeXtBackbone=MiT-B2×2, Params=58.8M, GFLOPs=74.2, FPS=31.0, Modality=RGB + thermal2026.04 | 78.9 | |
| DeepLabV3Backbone=ResNet-50, Params=24.9M, GFLOPs=110.0, FPS=130.0, Modality=thermal only2026.04 | 67.1 | |
| SegFormerBackbone=MiT-B2, Params=24.9M, GFLOPs=25.3, FPS=66.2, Modality=thermal only2026.04 | 64.7 | |
| ConvNeXt-BBackbone=ConvNeXt-B, Params=88.5M, GFLOPs=85.8, FPS=59.0, Modality=thermal only2026.04 | 63 | |
| Swin-SBackbone=Swin-S, Params=49.6M, GFLOPs=54.5, FPS=40.4, Modality=thermal only2026.04 | 61.2 | |
| CMXBackbone=MiT-B2×2, Params=66.7M, GFLOPs=78.3, FPS=31.1, Modality=RGB + thermal2026.04 | 47.8 |