Object Detection on NWPU (AP, AP50, AP75)
70.3APRSPrompter (anchor)
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| RSPrompter (anchor)Backbone=ViT-Base, Params (M)=117.09, bb-Params (M)=89.67, bb-FLOPs (G)=9722026.06 | 70.3 | 93.6 | 81 | |
| SAM-seg (Mask RCNN)Backbone=ViT-Base, Params (M)=309.69, bb-Params (M)=89.67, bb-FLOPs (G)=9722026.06 | 68.8 | 92.2 | 80.1 | |
| RSPrompter (query)Backbone=ViT-Base, Params (M)=101.02, bb-Params (M)=89.67, bb-FLOPs (G)=9722026.06 | 68.4 | 90.3 | 74 | |
| Cascade Mask RCNNBackbone=ResNet50, Params (M)=77.02, bb-Params (M)=25.56, bb-FLOPs (G)=172.692026.06 | 67.5 | 92.4 | 77.3 | |
| BoxInstBackbone=ResNet50, Params (M)=35.14, bb-Params (M)=25.56, bb-FLOPs (G)=172.692026.06 | 64.8 | 89.3 | 73 | |
| SAM-detBackbone=ViT-Base, Params (M)=135.08, bb-Params (M)=89.67, bb-FLOPs (G)=9722026.06 | 64.2 | 89.6 | 74.6 | |
| HTCBackbone=ResNet50, Params (M)=46.59, bb-Params (M)=25.56, bb-FLOPs (G)=172.692026.06 | 63.9 | 88.9 | 75.4 | |
| HQ-ISNetBackbone=ResNet50, Params (M)=80.56, bb-Params (M)=25.56, bb-FLOPs (G)=172.692026.06 | 63.5 | 89.9 | 75 | |
| CATNetBackbone=ResNet50, Params (M)=54.77, bb-Params (M)=25.56, bb-FLOPs (G)=172.692026.06 | 63.2 | 89 | 73.8 | |
| SAM-seg (Mask2Former)Backbone=ViT-Base, Params (M)=112.16, bb-Params (M)=89.67, bb-FLOPs (G)=9722026.06 | 63.1 | 86.3 | 70.6 | |
| Mask RCNNBackbone=ResNet50, Params (M)=44.18, bb-Params (M)=25.56, bb-FLOPs (G)=172.692026.06 | 62.3 | 88.3 | 75.2 | |
| MS RCNNBackbone=ResNet50, Params (M)=60.74, bb-Params (M)=25.56, bb-FLOPs (G)=172.692026.06 | 62.3 | 88.6 | 73.1 | |
| CondInstBackbone=ResNet50, Params (M)=34.16, bb-Params (M)=25.56, bb-FLOPs (G)=172.692026.06 | 62.3 | 87.8 | 73.3 | |
| RS4D (box)Backbone=SSM, Params (M)=33.20, bb-Params (M)=11.75, bb-FLOPs (G)=108.82026.06 | 61.7 | 85.8 | 70.1 | |
| RS4D (prompt)Backbone=SSM, Params (M)=38.83, bb-Params (M)=11.75, bb-FLOPs (G)=108.82026.06 | 61.5 | 86.9 | 73.7 | |
| SCNetBackbone=ResNet50, Params (M)=94.80, bb-Params (M)=25.56, bb-FLOPs (G)=172.692026.06 | 60 | 87.5 | 69.1 | |
| Mask2FormerBackbone=ResNet50, Params (M)=44.02, bb-Params (M)=25.56, bb-FLOPs (G)=172.692026.06 | 57.4 | 75.5 | 63.7 | |
| ViTDetBackbone=Swin-Huge, Params (M)=679.42, bb-Params (M)=655.58, bb-FLOPs (G)=375.102026.06 | 54.8 | 85.4 | 64.7 | |
| SwinV2(HTC++)Backbone=SwinV2-Huge, Params (M)=709.09, bb-Params (M)=655.14, bb-FLOPs (G)=303.372026.06 | 43.5 | 79.5 | 43.4 | |
| CBNetv2Backbone=Swin-Tiny, Params (M)=75.58, bb-Params (M)=28.30, bb-FLOPs (G)=12.472026.06 | 41 | 70.3 | 42.4 | |
| SAM-clsBackbone=ViT-Base2026.06 | 40.2 | 57.1 | 44.5 |