Object Detection on MSCOCO 2017 (val)
47.5APbDefMamba-S
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| DefMamba-SBackbone=DefMamba-S, #FLOPS=268G, Framework=Mask R-CNN, Schedule=1x2025.04 | 47.5 | 69.6 | 51.7 | |
| VMamba-TBackbone=VMamba-T, #FLOPS=286G, Framework=Mask R-CNN, Schedule=1x2025.04 | 47.4 | 69.5 | 52 | |
| GrootV-TBackbone=GrootV-T, #FLOPS=265G, Framework=Mask R-CNN, Schedule=1x2025.04 | 47 | 69.4 | 51.5 | |
| MSVMamba-TBackbone=MSVMamba-T, #FLOPS=252G, Framework=Mask R-CNN, Schedule=1x2025.04 | 46.9 | 68.8 | 51.4 | |
| LocalVMamba-TBackbone=LocalVMamba-T, #FLOPS=291G, Framework=Mask R-CNN, Schedule=1x2025.04 | 46.7 | 68.7 | 50.8 | |
| QuadMamba-SBackbone=QuadMamba-S, #FLOPS=301G, Framework=Mask R-CNN, Schedule=1x2025.04 | 46.7 | 69 | 51.3 | |
| PVTv2-B2Backbone=PVTv2-B2, #FLOPS=309G, Framework=Mask R-CNN, Schedule=1x2025.04 | 45.3 | 67.1 | 49.4 | |
| ConvNeXt-TBackbone=ConvNeXt-T, #FLOPS=262G, Framework=Mask R-CNN, Schedule=1x2025.04 | 44.2 | 66.6 | 48.3 | |
| Swin-TBackbone=Swin-T, #FLOPS=267G, Framework=Mask R-CNN, Schedule=1x2025.04 | 42.7 | 65.2 | 46.8 | |
| ResNet-50Backbone=ResNet-50, #FLOPS=260G, Framework=Mask R-CNN, Schedule=1x2025.04 | 38.2 | 58.8 | 41.4 | |
| H-GPE-SBackbone=H-GPE-S, Detection Head=SSDLite, Params(M)=6.1, FLOPs(G)=3.22026.01 | 30.3 | — | — | |
| EMB-SBackbone=EMB-S, Detection Head=SSDLite, Params(M)=5.6, FLOPs(G)=3.42026.01 | 28.8 | — | — | |
| EdgeNeXt-SBackbone=EdgeNeXt-S, Detection Head=SSDLite, Params(M)=6.2, FLOPs(G)=2.12026.01 | 27.9 | — | — | |
| EMO-5MBackbone=EMO-5M, Detection Head=SSDLite, Params(M)=6, FLOPs(G)=1.82026.01 | 27.9 | — | — | |
| MobileViTv2-1.25Backbone=MobileViTv2-1.25, Detection Head=SSDLite, Params(M)=8.2, FLOPs(G)=4.72026.01 | 27.8 | — | — | |
| MobileViTv1-SBackbone=MobileViTv1-S, Detection Head=SSDLite, Params(M)=5.7, FLOPs(G)=3.42026.01 | 27.7 | — | — | |
| PartialFormer-B1Backbone=PartialFormer-B1, Detection Head=SSDLite, Params(M)=8, FLOPs(G)=1.52026.01 | 27.1 | — | — | |
| H-GPE-TBackbone=H-GPE-T, Detection Head=SSDLite, Params(M)=2.9, FLOPs(G)=1.32026.01 | 26.8 | — | — | |
| EMB-TBackbone=EMB-T, Detection Head=SSDLite, Params(M)=2.7, FLOPs(G)=1.42026.01 | 25.7 | — | — | |
| MoConv-SBackbone=MoConv-S, Detection Head=SSDLite, Params(M)=3.7, FLOPs(G)=1.62026.01 | 25.3 | — | — | |
| ResNet50Backbone=ResNet50, Detection Head=SSDLite, Params(M)=26.6, FLOPs(G)=8.82026.01 | 25.2 | — | — | |
| EMO-2MBackbone=EMO-2M, Detection Head=SSDLite, Params(M)=3.3, FLOPs(G)=0.92026.01 | 25.2 | — | — | |
| MobileViTv2-0.75Backbone=MobileViTv2-0.75, Detection Head=SSDLite, Params(M)=3.6, FLOPs(G)=1.82026.01 | 24.6 | — | — | |
| PartialFormer-B0Backbone=PartialFormer-B0, Detection Head=SSDLite, Params(M)=5, FLOPs(G)=0.92026.01 | 24.3 | — | — | |
| H-GPE-NBackbone=H-GPE-N, Detection Head=SSDLite, Params(M)=1.9, FLOPs(G)=0.82026.01 | 24.1 | — | — | |
| MoConv-TBackbone=MoConv-T, Detection Head=SSDLite, Params(M)=2.5, FLOPs(G)=12026.01 | 23.8 | — | — | |
| EMB-NBackbone=EMB-N, Detection Head=SSDLite, Params(M)=1.7, FLOPs(G)=0.92026.01 | 23.6 | — | — | |
| MobileNetv1Backbone=MobileNetv1, Detection Head=SSDLite, Params(M)=5.1, FLOPs(G)=1.32026.01 | 22.2 | — | — | |
| MobileNetv2Backbone=MobileNetv2, Detection Head=SSDLite, Params(M)=4.3, FLOPs(G)=0.82026.01 | 22.1 | — | — | |
| MobileNetv3Backbone=MobileNetv3, Detection Head=SSDLite, Params(M)=5, FLOPs(G)=0.62026.01 | 22 | — | — | |
| EMO-1MBackbone=EMO-1M, Detection Head=SSDLite, Params(M)=2.3, FLOPs(G)=0.62026.01 | 22 | — | — | |
| MobileViTv2-0.5Backbone=MobileViTv2-0.5, Detection Head=SSDLite, Params(M)=2, FLOPs(G)=0.92026.01 | 21.2 | — | — | |
| MobileViTv1-XXSBackbone=MobileViTv1-XXS, Detection Head=SSDLite, Params(M)=1.7, FLOPs(G)=0.92026.01 | 19.9 | — | — |