Object Detection on MS COCO
73.3AP50InfoMamba-B
Evaluation results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| InfoMamba-BParams (M)=145, Detector=Cascade Mask R-CNN, Schedule=3×, Crop Size=1280 × 8002026.03 | 73.3 | 55.3 | — | — | — | — | — | — | — | — | 60.1 | — | |
| InfoMamba-SParams (M)=108, Detector=Cascade Mask R-CNN, Schedule=3×, Crop Size=1280 × 8002026.03 | 73.1 | 54.8 | — | — | — | — | — | — | — | — | 59.6 | — | |
| InfoMamba-TParams (M)=86, Detector=Cascade Mask R-CNN, Schedule=3×, Crop Size=1280 × 8002026.03 | 72 | 53.6 | — | — | — | — | — | — | — | — | 58.5 | — | |
| VMamba-BBackbone=VMamba-B, #Param=108M, FLOPs=485G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 71.4 | 49.2 | — | — | — | — | — | — | — | — | 54 | — | |
| ConvNeXt-BParams (M)=146, Detector=Cascade Mask R-CNN, Schedule=3×, Crop Size=1280 × 8002026.03 | 71.3 | 52.7 | — | — | — | — | — | — | — | — | 57.2 | — | |
| MambaVision-BParams (M)=145, Detector=Cascade Mask R-CNN, Schedule=3×, Crop Size=1280 × 8002026.03 | 71.3 | 52.8 | — | — | — | — | — | — | — | — | 57.2 | — | |
| MambaVision-SParams (M)=108, Detector=Cascade Mask R-CNN, Schedule=3×, Crop Size=1280 × 8002026.03 | 71.1 | 52.3 | — | — | — | — | — | — | — | — | 56.7 | — | |
| ConvNeXt-SParams (M)=108, Detector=Cascade Mask R-CNN, Schedule=3×, Crop Size=1280 × 8002026.03 | 70.8 | 51.9 | — | — | — | — | — | — | — | — | 56.5 | — | |
| LLMDetTraining Stage=-2026.05 | 70.8 | — | — | — | — | — | — | — | — | — | — | 54.4 | |
| LLMDetTrain Category=N/A2026.05 | 70.8 | 54.4 | — | — | — | — | — | — | — | — | — | — | |
| Swin-SParams (M)=107, Detector=Cascade Mask R-CNN, Schedule=3×, Crop Size=1280 × 8002026.03 | 70.7 | 51.9 | — | — | — | — | — | — | — | — | 56.3 | — | |
| MFilMamba-BBackbone=MFilMamba-B, #Param=112M, FLOPs=467G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 70.6 | 49 | — | — | — | — | — | — | — | — | 53.7 | — | |
| Swin-BParams (M)=145, Detector=Cascade Mask R-CNN, Schedule=3×, Crop Size=1280 × 8002026.03 | 70.5 | 51.9 | — | — | — | — | — | — | — | — | 56.4 | — | |
| MambaVision-TParams (M)=86, Detector=Cascade Mask R-CNN, Schedule=3×, Crop Size=1280 × 8002026.03 | 70 | 51.1 | — | — | — | — | — | — | — | — | 55.6 | — | |
| VMamba-SBackbone=VMamba-S, #Param=70M, FLOPs=349G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 70 | 48.7 | — | — | — | — | — | — | — | — | 53.4 | — | |
| DAT-SBackbone=DAT-S, #Param=69M, FLOPs=378G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 69.9 | 47.1 | — | — | — | — | — | — | — | — | 51.5 | — | |
| MFilMamba-SBackbone=MFilMamba-S, #Param=70M, FLOPs=320G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 69.4 | 47.9 | — | — | — | — | — | — | — | — | 52.3 | — | |
| ConvNeXt-BBackbone=ConvNeXt-B, #Param=108M, FLOPs=486G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 69.4 | 47 | — | — | — | — | — | — | — | — | 51.7 | — | |
| VMamba-TBackbone=VMamba-T, #Param=50M, FLOPs=271G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 69.3 | 47.3 | — | — | — | — | — | — | — | — | 52 | — | |
| MambaOut-BaseBackbone=MambaOut-Base, #Param=100M, FLOPs=495G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 69.3 | 47.4 | — | — | — | — | — | — | — | — | 52.2 | — | |
| Swin-TParams (M)=86, Detector=Cascade Mask R-CNN, Schedule=3×, Crop Size=1280 × 8002026.03 | 69.2 | 50.4 | — | — | — | — | — | — | — | — | 54.7 | — | |
| MFilMamba-TBackbone=MFilMamba-T, #Param=53M, FLOPs=261G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 69.2 | 47.3 | — | — | — | — | — | — | — | — | 51.9 | — | |
| ConvNeXt-TParams (M)=86, Detector=Cascade Mask R-CNN, Schedule=3×, Crop Size=1280 × 8002026.03 | 69.1 | 50.4 | — | — | — | — | — | — | — | — | 54.8 | — | |
| MambaOut-SmallBackbone=MambaOut-Small, #Param=65M, FLOPs=354G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 69.1 | 47.4 | — | — | — | — | — | — | — | — | 52.4 | — | |
| FractalMamba-TBackbone=FractalMamba-T, #Param=41M, FLOPs=266G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 68.7 | 46.8 | — | — | — | — | — | — | — | — | 50.8 | — | |
| ViT-Adapter-BBackbone=ViT-Adapter-B, #Param=120M, FLOPs=557G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 68.2 | 47 | — | — | — | — | — | — | — | — | 51.4 | — | |
| PlainMamba-L3Backbone=PlainMamba-L3, #Param=79M, FLOPs=696G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 68 | 46.8 | — | — | — | — | — | — | — | — | 51.1 | — | |
| ConvNeXt-SBackbone=ConvNeXt-S, #Param=70M, FLOPs=348G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 67.9 | 45.4 | — | — | — | — | — | — | — | — | 50 | — | |
| DAT-TBackbone=DAT-T, #Param=48M, FLOPs=272G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 67.6 | 44.4 | — | — | — | — | — | — | — | — | 48.5 | — | |
| KeepLoRaTraining Stage=Animals-252026.05 | 67.6 | — | — | — | — | — | — | — | — | — | — | 50.2 | |
| MambaOut-TinyBackbone=MambaOut-Tiny, #Param=43M, FLOPs=262G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 67.3 | 45.1 | — | — | — | — | — | — | — | — | 49.6 | — | |
| DeiT-Small, 16×16 patchesParams (M)=80, Detector=Cascade Mask R-CNN, Schedule=3×, Crop Size=1280 × 8002026.03 | 67.2 | 48 | — | — | — | — | — | — | — | — | 51.7 | — | |
| PlainMamba-L2Backbone=PlainMamba-L2, #Param=53M, FLOPs=542G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 66.9 | 46 | — | — | — | — | — | — | — | — | 50.1 | — | |
| ConvNeXt-TBackbone=ConvNeXt-T, #Param=48M, FLOPs=262G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 66.6 | 44.2 | — | — | — | — | — | — | — | — | 48.3 | — | |
| Swin-SBackbone=Swin-S, #Param=69M, FLOPs=354G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 66.6 | 44.8 | — | — | — | — | — | — | — | — | 48.9 | — | |
| X101-32Params (M)=101, Detector=Cascade Mask R-CNN, Schedule=3×, Crop Size=1280 × 8002026.03 | 66.5 | 48.1 | — | — | — | — | — | — | — | — | 52.4 | — | |
| X101-64Params (M)=140, Detector=Cascade Mask R-CNN, Schedule=3×, Crop Size=1280 × 8002026.03 | 66.4 | 48.3 | — | — | — | — | — | — | — | — | 52.3 | — | |
| EffVMamba-BBackbone=EffVMamba-B, #Param=53M, FLOPs=252G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 66.2 | 43.7 | — | — | — | — | — | — | — | — | 47.9 | — | |
| ViT-Adapter-SBackbone=ViT-Adapter-S, #Param=48M, FLOPs=403G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 65.8 | 44.7 | — | — | — | — | — | — | — | — | 48.3 | — | |
| Swin-TBackbone=Swin-T, #Param=48M, FLOPs=267G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 65.2 | 42.7 | — | — | — | — | — | — | — | — | 46.8 | — | |
| NoIn-DetTraining Stage=Animals-252026.05 | 65.1 | — | — | — | — | — | — | — | — | — | — | 49.8 | |
| PVT-LBackbone=PVT-L, #Param=81M, FLOPs=364G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 65 | 42.9 | — | — | — | — | — | — | — | — | 46.6 | — | |
| PlainMamba-L1Backbone=PlainMamba-L1, #Param=31M, FLOPs=388G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 64.8 | 44.1 | — | — | — | — | — | — | — | — | 47.9 | — | |
| PVT-MBackbone=PVT-M, #Param=64M, FLOPs=302G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 64.4 | 42 | — | — | — | — | — | — | — | — | 45.6 | — | |
| ResNet-50Params (M)=82, Detector=Cascade Mask R-CNN, Schedule=3×, Crop Size=1280 × 8002026.03 | 64.3 | 46.3 | — | — | — | — | — | — | — | — | 50.5 | — | |
| NoIn-DetTraining Stage=Clothing-62026.05 | 64.3 | — | — | — | — | — | — | — | — | — | — | 49.5 | |
| C-CLIPTraining Stage=Animals-252026.05 | 64 | — | — | — | — | — | — | — | — | — | — | 48.3 | |
| NoIn-DetTraining Stage=Cold Weapons-42026.05 | 63.7 | — | — | — | — | — | — | — | — | — | — | 49.1 | |
| C-CLIPTraining Stage=Food-282026.05 | 63.1 | — | — | — | — | — | — | — | — | — | — | 45.4 | |
| PVT-SBackbone=PVT-S, #Param=44M, FLOPs=245G, Framework=Mask R-CNN, Training schedule=1×, Input resolution=1280 × 800, Pre-training=ImageNet-1K2026.03 | 62.9 | 40.4 | — | — | — | — | — | — | — | — | 43.8 | — | |
| C-CLIPTraining Stage=Instruments-82026.05 | 62.6 | — | — | — | — | — | — | — | — | — | — | 46.6 | |
| NoIn-DetTraining Stage=Food-282026.05 | 62 | — | — | — | — | — | — | — | — | — | — | 47 | |
| C-CLIPTrain Category=Instruments-82026.05 | 62 | 46.2 | — | — | — | — | — | — | — | — | — | — | |
| KeepLoRaTrain Category=Food-282026.05 | 62 | 46 | — | — | — | — | — | — | — | — | — | — | |
| NoIn-DetTrain Category=Animals-252026.05 | 62 | 46.6 | — | — | — | — | — | — | — | — | — | — | |
| NoIn-DetTraining Stage=Instruments-82026.05 | 61.9 | — | — | — | — | — | — | — | — | — | — | 47.1 | |
| C-CLIPTrain Category=Cold Weapons-42026.05 | 61.6 | 45.5 | — | — | — | — | — | — | — | — | — | — | |
| C-CLIPTraining Stage=Cold Weapons-42026.05 | 61.5 | — | — | — | — | — | — | — | — | — | — | 45.2 | |
| SwAV+UOTAFramework=Faster R-CNN, Backbone=ResNet-50, FPN=true, Finetuning schedule=1x, Pre-training epochs=2002021.12 | 61 | 39 | — | — | — | — | — | — | — | — | 42 | — | |
| OursBias strategy=Frequency Aware2025.10 | 61 | 40.3 | — | — | — | — | — | 43.8 | 52.5 | — | 44 | — | |
| C-CLIPTrain Category=Food-282026.05 | 61 | 45.6 | — | — | — | — | — | — | — | — | — | — | |
| ZiRaTraining Stage=Animals-252026.05 | 60.7 | — | — | — | — | — | — | — | — | — | — | 46.4 | |
| SwAVFramework=Faster R-CNN, Backbone=ResNet-50, FPN=true, Finetuning schedule=1x, Pre-training epochs=2002021.12 | 60.5 | 38.5 | — | — | — | — | — | — | — | — | 41.4 | — | |
| FPNBackbone=ResNet-101, IoU-guided NMS=false, Bounding Box Refinement=false2018.07 | 60.3 | 38.5 | 55.5 | 47.6 | 33.8 | 11.3 | — | — | — | — | — | — | |
| KeepLoRaTraining Stage=Food-282026.05 | 60.3 | — | — | — | — | — | — | — | — | — | — | 45.1 | |
| IoU-NetBackbone=ResNet-101, IoU-guided NMS=true, Bounding Box Refinement=false2018.07 | 60.2 | 38.9 | 55.5 | 47.8 | 34.6 | 12 | — | — | — | — | — | — | |
| NoIn-DetTraining Stage=Plants-252026.05 | 60.2 | — | — | — | — | — | — | — | — | — | — | 45.7 | |
| Whole DatasetIPD=100%, Student model=Faster R-CNN-502025.06 | 60.1 | 39.8 | — | — | — | — | — | — | — | — | 43.3 | — | |
| KeepLoRaTraining Stage=Plants-252026.05 | 59.9 | — | — | — | — | — | — | — | — | — | — | 44.7 | |
| KeepLoRaTraining Stage=Vessels-182026.05 | 59.7 | — | — | — | — | — | — | — | — | — | — | 43.8 | |
| NoIn-DetTrain Category=Cold Weapons-42026.05 | 59.7 | 45.2 | — | — | — | — | — | — | — | — | — | — | |
| supervisedFramework=Faster R-CNN, Backbone=ResNet-50, FPN=true, Finetuning schedule=1x, Pre-training epochs=N/A2021.12 | 59.1 | 38.2 | — | — | — | — | — | — | — | — | 41.5 | — | |
| NoIn-DetTrain Category=Instruments-82026.05 | 59.1 | 44.9 | — | — | — | — | — | — | — | — | — | — | |
| IoU-NetBackbone=ResNet-101, IoU-guided NMS=false, Bounding Box Refinement=true2018.07 | 59 | 40 | 55.1 | 48.6 | 37 | 15.5 | — | — | — | — | — | — | |
| IoU-NetBackbone=ResNet-101, IoU-guided NMS=true, Bounding Box Refinement=true2018.07 | 59 | 40.6 | 55.2 | 49 | 38 | 17.1 | — | — | — | — | — | — | |
| KeepLoRaTrain Category=Animals-252026.05 | 58.9 | 43.6 | — | — | — | — | — | — | — | — | — | — | |
| NoIn-DetTrain Category=Food-282026.05 | 58.7 | 44.8 | — | — | — | — | — | — | — | — | — | — | |
| GeoDiffusionBias strategy=Frequency Aware, Resampling=true2025.10 | 58.6 | 38.5 | — | — | — | — | — | 42.2 | 49.9 | — | 42.4 | — | |
| GeoDiffusionBias strategy=Bias Agnostic, Resampling=false2025.10 | 58.5 | 38.4 | — | — | — | — | — | 42.1 | 50.3 | — | 42.4 | — | |
| IoU-NetBackbone=ResNet-50, IoU-guided NMS=true, Bounding Box Refinement=false2018.07 | 58.3 | 37 | 53.8 | 45.7 | 31.9 | 10.7 | — | — | — | — | — | — | |
| JCLFramework=Faster R-CNN, Backbone=ResNet-50, FPN=true, Finetuning schedule=1x, Pre-training epochs=2002021.12 | 58.3 | 38.1 | — | — | — | — | — | — | — | — | 41.3 | — | |
| Faster R-CNNstatus=Baseline2025.10 | 58.1 | 37.4 | — | — | — | — | — | 41 | 48.1 | — | 40.4 | — | |
| FPNBackbone=ResNet-50, IoU-guided NMS=false, Bounding Box Refinement=false2018.07 | 58 | 36.4 | 53.1 | 44.9 | 31.2 | 9.8 | — | — | — | — | — | — | |
| C-CLIPTraining Stage=Clothing-62026.05 | 58 | — | — | — | — | — | — | — | — | — | — | 43.3 | |
| MoCo-v2Framework=Faster R-CNN, Backbone=ResNet-50, FPN=true, Finetuning schedule=1x, Pre-training epochs=2002021.12 | 57.9 | 37.6 | — | — | — | — | — | — | — | — | 40.8 | — | |
| ControlNetBias strategy=Bias Agnostic, Resampling=false2025.10 | 57.8 | 36.9 | — | — | — | — | — | 40.4 | 49 | — | 39.6 | — | |
| ControlNetBias strategy=Frequency Aware, Resampling=true2025.10 | 57.8 | 36.9 | — | — | — | — | — | 40.5 | 47.6 | — | 39.7 | — | |
| GLIGENBias strategy=Bias Agnostic, Resampling=false2025.10 | 57.6 | 36.8 | — | — | — | — | — | 40.3 | 47.9 | — | 39.9 | — | |
| MoCo-v1Framework=Faster R-CNN, Backbone=ResNet-50, FPN=true, Finetuning schedule=1x, Pre-training epochs=2002021.12 | 57.4 | 37.1 | — | — | — | — | — | — | — | — | 40.2 | — | |
| L.DiffuseBias strategy=Bias Agnostic, Resampling=false2025.10 | 57.4 | 36.6 | — | — | — | — | — | 40 | 47.4 | — | 39.5 | — | |
| ZiRaTraining Stage=Clothing-62026.05 | 57.2 | — | — | — | — | — | — | — | — | — | — | 43.7 | |
| C-CLIPTrain Category=Animals-252026.05 | 57.2 | 42.7 | — | — | — | — | — | — | — | — | — | — | |
| L.DiffusionBias strategy=Bias Agnostic, Resampling=false2025.10 | 57 | 36.5 | — | — | — | — | — | 39.7 | 47.5 | — | 39.5 | — | |
| KeepLoRaTrain Category=Clothing-62026.05 | 56.9 | 40.6 | — | — | — | — | — | — | — | — | — | — | |
| ZiRaTraining Stage=Cold Weapons-42026.05 | 56.4 | — | — | — | — | — | — | — | — | — | — | 43 | |
| IoU-NetBackbone=ResNet-50, IoU-guided NMS=true, Bounding Box Refinement=true2018.07 | 56.3 | 38.1 | 52.4 | 46.3 | 35.1 | 15.5 | — | — | — | — | — | — | |
| NoIn-DetTrain Category=Clothing-62026.05 | 56.3 | 42.5 | — | — | — | — | — | — | — | — | — | — | |
| IoU-NetBackbone=ResNet-50, IoU-guided NMS=false, Bounding Box Refinement=true2018.07 | 56.2 | 37.6 | 52.4 | 46 | 34.1 | 14 | — | — | — | — | — | — | |
| KeepLoRaTrain Category=Instruments-82026.05 | 56 | 40 | — | — | — | — | — | — | — | — | — | — | |
| C-CLIPTrain Category=Clothing-62026.05 | 55.9 | 41.9 | — | — | — | — | — | — | — | — | — | — |