Object Detection on MSCOCO (val)
61.3APHIPIE
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| HIPIEBackbone=ViT-H2023.07 | 61.3 | — | — | — | 45.8 | 65.7 | 75.9 | |
| ViTDet-HBackbone=ViT-H2023.07 | 58.7 | — | — | — | — | — | — | |
| UNINEXTBackbone=ViT-H2023.07 | 58.1 | — | — | — | 40.7 | 62.5 | 73.6 | |
| Cas. Mask-RCNNBackbone=CNeXtL2023.07 | 54.8 | — | — | — | — | — | — | |
| HIPIEBackbone=RN502023.07 | 53.9 | — | — | — | 37.5 | 58 | 68 | |
| YOLOv6-LSize=640, Latency(ms)=9.87, GFLOPs=144, Params(M)=58.5, Distillation=true2022.11 | 52.5 | — | 70 | — | — | — | — | |
| DAMO-YOLO-LSize=640, Latency(ms)=7.95, GFLOPs=97.3, Params(M)=42.1, FPSV100=126, Distillation=true2022.11 | 51.9 | — | 68.5 | 56.7 | 33.3 | 57 | 67.6 | |
| YOLOE-LSize=640, Latency(ms)=9.94, GFLOPs=110.1, Params(M)=52.2, FPSV100=149, Distillation=false2022.11 | 51.4 | — | 68.9 | 55.6 | 31.4 | 55.3 | 66.1 | |
| UNINEXTBackbone=RN502023.07 | 51.3 | — | — | — | 32.6 | 55.7 | 66.5 | |
| YOLOv7Size=640, Latency(ms)=9.08, GFLOPs=104.7, Params(M)=36.9, Distillation=false2022.11 | 51.2 | — | 69.7 | 55.9 | 31.8 | 55.5 | 65 | |
| DAMO-YOLO-LSize=640, Latency(ms)=7.95, GFLOPs=97.3, Params(M)=42.1, FPSV100=126, Distillation=false2022.11 | 50.8 | — | 67.5 | 55.5 | 33.2 | 55.7 | 66.6 | |
| DAMO-YOLO-MSize=640, Latency(ms)=5.62, GFLOPs=61.8, Params(M)=28.2, FPSV100=233, Distillation=true2022.11 | 50.4 | — | 67.2 | 55.1 | 31.6 | 55.3 | 67.1 | |
| YOLOX-LSize=640, Latency(ms)=11.44, GFLOPs=155.6, Params(M)=54.2, FPSV100=120, Distillation=false2022.11 | 49.7 | — | — | — | — | — | — | |
| YOLOv6-MSize=640, Latency(ms)=5.72, GFLOPs=82.2, Params(M)=34.3, Distillation=true2022.11 | 49.5 | — | 66.8 | — | — | — | — | |
| DAMO-YOLO-MSize=640, Latency(ms)=5.62, GFLOPs=61.8, Params(M)=28.2, FPSV100=233, Distillation=false2022.11 | 49.2 | — | 65.5 | 53 | 29.7 | 53.1 | 66.1 | |
| YOLOv5-LSize=640, Latency(ms)=8.92, GFLOPs=109.1, Params(M)=46.5, FPSV100=172, Distillation=false2022.11 | 49 | — | 67.3 | — | — | — | — | |
| YOLOE-MSize=640, Latency(ms)=6.67, GFLOPs=49.9, Params(M)=23.4, FPSV100=208, Distillation=false2022.11 | 49 | — | 66.5 | 53 | 28.6 | 52.9 | 63.8 | |
| DN-DETRBackbone=RN502023.07 | 48.6 | — | — | — | 31 | 52 | 63.7 | |
| DAMO-YOLO-SSize=640, Latency(ms)=3.83, GFLOPs=37.8, Params(M)=16.3, FPSV100=325, Distillation=true2022.11 | 47.7 | — | 63.5 | 51.1 | 26.9 | 51.7 | 64.9 | |
| YOLOX-MSize=640, Latency(ms)=6.46, GFLOPs=73.8, Params(M)=25.3, FPSV100=177, Distillation=false2022.11 | 46.9 | — | — | — | — | — | — | |
| Deform. DETRBackbone=RN502023.07 | 46.9 | — | — | — | 29.6 | 50.1 | 61.6 | |
| DAMO-YOLO-SSize=640, Latency(ms)=3.83, GFLOPs=37.8, Params(M)=16.3, FPSV100=325, Distillation=false2022.11 | 46 | — | 61.9 | 49.5 | 25.9 | 50.6 | 62.5 | |
| YOLOv5-MSize=640, Latency(ms)=5.71, GFLOPs=49, Params(M)=21.2, FPSV100=263, Distillation=false2022.11 | 45.4 | — | 64.1 | — | — | — | — | |
| DAMO-YOLO-TSize=640, Latency(ms)=2.78, GFLOPs=18.1, Params(M)=8.5, FPSV100=397, Distillation=true2022.11 | 43.6 | — | 59.4 | 46.6 | 23.3 | 47.4 | 61 | |
| YOLOv6-SSize=640, Latency(ms)=3.1, GFLOPs=44.2, Params(M)=17, Distillation=false2022.11 | 43.5 | — | 60.4 | — | — | — | — | |
| YOLOE-SSize=640, Latency(ms)=3.21, GFLOPs=17.4, Params(M)=7.9, FPSV100=333, Distillation=false2022.11 | 43 | — | 60.5 | 46.6 | 23.2 | 46.4 | 56.9 | |
| DAMO-YOLO-TSize=640, Latency(ms)=2.78, GFLOPs=18.1, Params(M)=8.5, FPSV100=397, Distillation=false2022.11 | 42 | — | 58 | 45.2 | 23 | 46.1 | 58.5 | |
| YOLOX-SSize=640, Latency(ms)=3.2, GFLOPs=26.8, Params(M)=9, FPSV100=247, Distillation=false2022.11 | 40.5 | — | — | — | — | — | — | |
| DAMO-YOLO-NlSize=416, Latency(ms)=6.69, GFLOPs=6.04, Params(M)=5.692022.11 | 40.5 | — | 57.6 | 43.3 | 20.7 | 44.5 | 60.9 | |
| YOLOv6-TSize=640, Latency(ms)=2.53, GFLOPs=36.7, Params(M)=15, Distillation=false2022.11 | 40.3 | — | 56.6 | — | — | — | — | |
| Picodet-lSize=416, Latency(ms)=7.03, GFLOPs=7.10, Params(M)=5.802022.11 | 39.4 | — | — | — | — | — | — | |
| YOLOv7-T-siluSize=640, Latency(ms)=3.13, GFLOPs=13.7, Params(M)=6.2, Distillation=false2022.11 | 38.7 | — | 56.7 | 41.7 | 18.8 | 42.4 | 51.9 | |
| DAMO-YOLO-NmSize=416, Latency(ms)=5.05, GFLOPs=3.69, Params(M)=2.142022.11 | 38.2 | — | 54.7 | 40.8 | 20.2 | 41.7 | 57.6 | |
| Picodet-mSize=416, Latency(ms)=5.69, GFLOPs=4.34, Params(M)=3.462022.11 | 37.5 | — | — | — | — | — | — | |
| YOLOv5-SSize=640, Latency(ms)=3.04, GFLOPs=16.5, Params(M)=7.2, FPSV100=455, Distillation=false2022.11 | 37.4 | — | 56.8 | — | — | — | — | |
| yolov8nSize=640, Latency(ms)=6.16, GFLOPs=8.7, Params(M)=3.22022.11 | 37.3 | — | — | — | — | — | — | |
| Nanodet-plus-m1.5xSize=416, Latency(ms)=6.09, GFLOPs=2.97, Params(M)=2.442022.11 | 34.1 | — | — | — | — | — | — | |
| YOLOX-TSize=416, Latency(ms)=1.78, GFLOPs=6.5, Params(M)=5.1, Distillation=false2022.11 | 32.8 | — | — | — | — | — | — | |
| Picodet-sSize=416, Latency(ms)=4.53, GFLOPs=1.65, Params(M)=1.182022.11 | 32.5 | — | — | — | — | — | — | |
| DAMO-YOLO-NsSize=416, Latency(ms)=4.08, GFLOPs=1.56, Params(M)=1.412022.11 | 32.3 | — | 47.7 | 34.2 | 12.3 | 34.8 | 51.8 | |
| Nanodet-plus-mSize=416, Latency(ms)=5.80, GFLOPs=1.52, Params(M)=1.172022.11 | 30.4 | — | — | — | — | — | — | |
| YOLOv5-NSize=640, Latency(ms)=2.23, GFLOPs=4.5, Params(M)=1.9, Distillation=false2022.11 | 28 | — | 45.7 | — | — | — | — | |
| Picodet-xsSize=416, Latency(ms)=4.27, GFLOPs=1.13, Params(M)=0.702022.11 | 26.2 | — | — | — | — | — | — | |
| Base (Swin-Base)Backbone=Swin-Base, Detector=Cascade Mask RCNN, Params(M)=145.0, GFLOPS=1,5012023.10 | — | 50.1 | — | — | — | — | — | |
| Cascade Eff-B7 NAS-FPN (1280)Backbone=EfficientNet-B7, Resolution=1280, Architecture=Cascade NAS-FPN, Test-time augmentation=false2022.04 | — | 54.8 | — | — | — | — | — | |
| Cascade Eff-B7 NAS-FPN (1280) with Copy-PasteBackbone=EfficientNet-B7, Resolution=1280, Architecture=Cascade NAS-FPN, Data Augmentation=Copy-Paste, Test-time augmentation=false2022.04 | — | 55.9 | — | — | — | — | — | |
| Cascade Eff-B7 NAS-FPN (1280) with Copy-Paste and frozen backboneBackbone=EfficientNet-B7, Resolution=1280, Architecture=Cascade NAS-FPN, Data Augmentation=Copy-Paste, Backbone freezing=true, Test-time augmentation=false2022.04 | — | 57 | — | — | — | — | — | |
| Cascade Eff-B7 NAS-FPN (1280) with frozen backboneBackbone=EfficientNet-B7, Resolution=1280, Architecture=Cascade NAS-FPN, Backbone freezing=true, Test-time augmentation=false2022.04 | — | 55.7 | — | — | — | — | — | |
| LORA (Swin-Base)Backbone=Swin-Base, Detector=Cascade Mask RCNN, Params(M)=149.0, GFLOPS=1,5472023.10 | — | 46.1 | — | — | — | — | — | |
| MaskAlign (Swin-Base)Backbone=Swin-Base, Detector=Cascade Mask RCNN, Params(M)=115.8, GFLOPS=1,3022023.10 | — | 41.1 | — | — | — | — | — | |
| PELA (Swin-Base)Backbone=Swin-Base, Detector=Cascade Mask RCNN, Params(M)=103.0, GFLOPS=1,2322023.10 | — | 49 | — | — | — | — | — | |
| ResNet-50Backbone=ResNet-50, Detector=Cascade Mask RCNN, Params(M)=77.3, GFLOPS=411.02023.10 | — | 46.3 | — | — | — | — | — | |
| ResNeXt-101-32Backbone=ResNeXt-101-32, Detector=Cascade Mask RCNN, Params(M)=96.0, GFLOPS=546.12023.10 | — | 48.1 | — | — | — | — | — | |
| Self-training Copy-PasteTraining Strategy=Self-training, Data Augmentation=Copy-Paste, Test-time augmentation=false2022.04 | — | 57 | — | — | — | — | — | |
| Soft Teacher + Swin-LBackbone=Swin-L, Test-time augmentation=false2022.04 | — | 59.1 | — | — | — | — | — | |
| Swin TransformerPre-training=ImageNet-22K, Test-time augmentation=false2022.04 | — | 57.1 | — | — | — | — | — | |
| TinyBert (Swin-Base)Backbone=Swin-Base, Detector=Cascade Mask RCNN, Params(M)=115.8, GFLOPS=1,3022023.10 | — | 41.1 | — | — | — | — | — |