Object Detection on PASCAL VOC 2007 (test)
85.8mAPRefineDet512+
Evaluation results
| Method | Links | ||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| RefineDet512+Backbone=VGG-16, Stage=one-stage, Multi-scale testing=true2017.11 | 85.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RefineDet320+Backbone=VGG-16, Stage=one-stage, Multi-scale testing=true2017.11 | 85.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Faster R-CNN +++training data=07+12+COCO, test time (sec/img)=3.36, Backbone=ResNet-101, iterative box regression=true, context=true, multi-scale testing=true2016.05 | 85.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RefineDet512Backbone=VGG-16, Stage=one-stage, Multi-scale testing=false2017.11 | 85.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RefineDet320Backbone=VGG-16, Stage=one-stage, Multi-scale testing=false2017.11 | 84 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SGDFBackbone=ResNet50, Framework=Faster-RCNN, Neck=FPN, Pre-training=COCO dataset2026.03 | 83.81 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RefineDet512+Backbone=VGG-162017.11 | 83.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| R-FCNBackbone=ResNet-101, Stage=two-stage, Multi-scale testing=false2017.11 | 83.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| R-FCN multi-sc traintraining data=07+12+COCO, test time (sec/img)=0.17, Backbone=ResNet-1012016.05 | 83.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SSD512Backbone=VGG-16, Stage=one-stage, Multi-scale testing=false2017.11 | 83.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RefineDet320+Backbone=VGG-162017.11 | 83.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CoupleNet context multi-sc traintraining data=07+12, GPU=TITAN X, test time (ms/img)=122, Backbone=ResNet-101, Input resolution=600x1000, context prior=true, multi-scale training=true2017.08 | 82.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CoupleNetTrain=07+12, Base Network=ResNet-1012017.08 | 82.7 | 85.7 | 87 | 84.8 | 75.5 | 73.3 | 88.8 | 89.2 | 89.6 | 69.8 | 87.5 | 76.1 | 88.9 | 89 | 87.2 | 86.2 | 59.1 | 83.6 | 83.4 | 87.6 | 80.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CoupleNetBackbone=ResNet-101, Input size=~1000 × 600, #Boxes=300, FPS=8.22017.11 | 82.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| TripleNet512backbone=ResNet1012018.09 | 82.7 | 88.9 | 87.8 | 83.7 | 79.6 | 62.9 | 87.9 | 88.3 | 88.5 | 67.5 | 89.1 | 81.2 | 88 | 89.5 | 87.9 | 83.3 | 58.7 | 85.1 | 83.4 | 88.8 | 84.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| InterNetStructure=ResNet-101, Training data=07+122019.03 | 82.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DFPR512backbone=ResNet1012018.09 | 82.4 | 92 | 88.2 | 81.1 | 71.2 | 65.7 | 88.2 | 87.9 | 92.2 | 65.8 | 86.5 | 79.4 | 90.3 | 90.4 | 89.3 | 88.6 | 59.4 | 88.4 | 75.3 | 89.2 | 78.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| TripleNet512backbone=ResNet502018.09 | 82.4 | 88.9 | 86.9 | 85 | 77.5 | 61.5 | 87.7 | 88.2 | 89.2 | 66 | 88.3 | 79.6 | 87.5 | 88.8 | 87.3 | 82.3 | 62.4 | 86.1 | 81.7 | 89.2 | 82.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Faster R-CNNBackbone=FPN, Initialization=ResNet-50 + MoEx2020.02 | 82.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 54.3 | — | — | — | — | |
| CoupleNet contexttraining data=07+12, GPU=TITAN X, test time (ms/img)=122, Backbone=ResNet-101, Input resolution=600x1000, context prior=true2017.08 | 82.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Faster R-CNNBackbone=C4, Initialization=ResNet-50 + CutMix2020.02 | 82.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 54.3 | — | — | — | — | |
| Faster R-CNNBackbone=FPN, Initialization=ResNet-50 + CutMix2020.02 | 82.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 54.3 | — | — | — | — | |
| Faster R-CNNBackbone=FPN, Initialization=ResNet-50 (300 epochs)2020.02 | 82 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 54.2 | — | — | — | — | |
| RefineDet512Backbone=VGG-16, Input size=512 x 512, #Boxes=16320, FPS=24.12017.11 | 81.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RefineDet512backbone=VGG162018.09 | 81.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Faster R-CNNBackbone=FPN, Initialization=ResNet-50 (default)2020.02 | 81.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 53.8 | — | — | — | — | |
| CoupleNettraining data=07+12, GPU=TITAN X, test time (ms/img)=102, Backbone=ResNet-101, Input resolution=600x10002017.08 | 81.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DSOD300Backbone=DS/64-192-48-1, Stage=one-stage, Multi-scale testing=false2017.11 | 81.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DES512backbone=VGG162018.09 | 81.7 | 87.7 | 86.7 | 85.2 | 76.3 | 60.6 | 88.7 | 89 | 88 | 67 | 86.9 | 78 | 87.2 | 87.9 | 86.7 | 84.4 | 59.2 | 86.1 | 79.2 | 88.1 | 80.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SSD512Training Data=COCO trainval35k then VOC07+12, Input Size=512x5122015.12 | 81.6 | 86.6 | 88.3 | 82.4 | 76 | 66.3 | 88.6 | 88.9 | 89.1 | 65.1 | 88.4 | 73.6 | 86.5 | 88.9 | 85.3 | 84.6 | 59.1 | 85 | 80.4 | 87.4 | 81.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Faster R-CNNBackbone=C4, Initialization=ResNet-50 + MoEx2020.02 | 81.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 54.6 | — | — | — | — | |
| DSSD513Backbone=ResNet-101, Input size=513 x 513, #Boxes=43688, FPS=5.52017.11 | 81.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DSSD513backbone=ResNet1012018.09 | 81.5 | 86.6 | 86.2 | 82.6 | 74.9 | 62.5 | 89 | 88.7 | 88.8 | 65.2 | 87 | 78.7 | 88.2 | 89 | 87.5 | 83.7 | 51.1 | 86.3 | 81.6 | 85.7 | 83.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| BlitzNet512backbone=ResNet502018.09 | 81.5 | 87 | 87.6 | 83.5 | 75.7 | 59.1 | 87.6 | 88 | 88.8 | 64.1 | 88.4 | 80.9 | 87.5 | 88.5 | 87.5 | 81.5 | 60.6 | 86.5 | 79.3 | 87.5 | 81.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SSD512Backbone=HarDNet-682019.09 | 81.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DPNetBackbone=DPNet, Input Size=320 x 320, FLOPS (G)=1, Params (M)=2.5, FPS=196, Pre-training=ImageNet 21K2022.09 | 81.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RON384++Backbone=VGG-16, Stage=one-stage, Multi-scale testing=false2017.11 | 81.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SSD300Backbone=VGG-16, Stage=one-stage, Multi-scale testing=false2017.11 | 81.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Faster R-CNNBackbone=C4, Initialization=ResNet-50 (300 epochs)2020.02 | 81.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 53.5 | — | — | — | — | |
| AdaBeliefBackbone=ResNet50, Framework=Faster-RCNN, Neck=FPN, Pre-training=COCO dataset2026.03 | 81.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| STDN513backbone=DenseNet1692018.09 | 80.9 | 86.1 | 89.3 | 79.5 | 74.3 | 61.9 | 88.5 | 88.3 | 89.4 | 67.4 | 86.5 | 79.5 | 86.4 | 89.2 | 88.5 | 79.3 | 53 | 77.9 | 81.4 | 86.6 | 85.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| EAdamBackbone=ResNet50, Framework=Faster-RCNN, Neck=FPN, Pre-training=COCO dataset2026.03 | 80.62 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SSD513Backbone=ResNet-101, Input size=513 x 513, #Boxes=43688, FPS=6.82017.11 | 80.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SSD512Backbone=ResNet-1012019.09 | 80.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| R-FCN multi-sc traintraining data=07+12, GPU=TITAN X, test time (ms/img)=83, Backbone=ResNet-101, Input resolution=600x1000, multi-scale training=true2017.08 | 80.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| R-FCNTrain=07+12, Base Network=ResNet-1012017.08 | 80.5 | 79.9 | 87.2 | 81.5 | 72 | 69.8 | 86.8 | 88.5 | 89.8 | 67 | 88.1 | 74.5 | 89.8 | 90.6 | 79.9 | 81.2 | 53.7 | 81.8 | 81.5 | 85.9 | 79.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| R-FCNBackbone=ResNet-101, Input size=~1000 × 600, #Boxes=300, FPS=92017.11 | 80.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| R-FCNStructure=ResNet-101, Training data=07+122019.03 | 80.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| R-FCN multi-sc traintraining data=07+12, test time (sec/img)=0.17, Backbone=ResNet-1012016.05 | 80.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SGDBackbone=ResNet50, Framework=Faster-RCNN, Neck=FPN, Pre-training=COCO dataset2026.03 | 80.43 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Faster R-CNNBackbone=C4, Initialization=ResNet-50 (default)2020.02 | 80.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 51.8 | — | — | — | — | |
| SSD512Implementations=ChainerCV, batch size=242017.08 | 80.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DPNetBackbone=DPNet, Input Size=320 x 320, FLOPS (G)=1, Params (M)=2.5, FPS=196, Pre-training=ImageNet 1K2022.09 | 80.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RefineDet320Backbone=VGG-16, Input size=320 × 320, #Boxes=6375, FPS=40.32017.11 | 80 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RefineDet320backbone=VGG162018.09 | 80 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SSD512*Backbone=VGG-16, Input size=512 x 512, #Boxes=24564, FPS=192017.11 | 79.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SSD512Backbone=VGG-162019.09 | 79.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DES300backbone=VGG162018.09 | 79.7 | 83.5 | 86 | 78.1 | 74.8 | 53.4 | 87.9 | 87.3 | 88.6 | 64 | 83.8 | 77.2 | 85.9 | 88.6 | 87.4 | 80.8 | 57.3 | 80.2 | 80.4 | 88.5 | 79.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SSD300Training Data=COCO trainval35k then VOC07+12, Input Size=300x3002015.12 | 79.6 | 80.9 | 86.3 | 79 | 76.2 | 57.6 | 87.3 | 88.2 | 88.6 | 60.5 | 85.4 | 76.7 | 87.5 | 89.2 | 84.5 | 81.4 | 55 | 81.9 | 81.5 | 85.9 | 78.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| R-FCNtraining data=07+12, GPU=TITAN X, test time (ms/img)=83, Backbone=ResNet-101, Input resolution=600x10002017.08 | 79.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SSD512*Train=07+12, Base Network=VGG162017.08 | 79.5 | 84.8 | 85.1 | 81.5 | 73 | 57.8 | 87.8 | 88.3 | 87.4 | 63.5 | 85.4 | 73.2 | 86.2 | 86.7 | 83.9 | 82.5 | 55.6 | 81.7 | 79 | 86.6 | 80 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SSD512Implementations=Original2017.08 | 79.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SSD512backbone=VGG162018.09 | 79.5 | 84.8 | 85.1 | 81.5 | 73 | 57.8 | 87.8 | 88.3 | 87.4 | 63.5 | 85.4 | 73.2 | 86.2 | 86.7 | 83.9 | 82.5 | 55.6 | 81.7 | 79 | 86.6 | 80 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Faster R-CNNBackbone=Conv R50, Neck=Conv2021.12 | 79.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| R-FCNtraining data=07+12, test time (sec/img)=0.17, Backbone=ResNet-1012016.05 | 79.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| StairNet + CBAMBackbone=VGG16, Detector=StairNet + CBAM, Parameters (M)=32.12018.07 | 79.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| TripleNet300backbone=ResNet502018.09 | 79.3 | 81.4 | 85 | 79.5 | 72.1 | 53.7 | 85.3 | 85.9 | 87.8 | 62.5 | 85.1 | 78.7 | 87.8 | 88.6 | 85.7 | 79.5 | 56.8 | 80.7 | 79.2 | 88.7 | 81.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DPNetBackbone=DPNet, Input Size=320 x 320, FLOPS (G)=1, Params (M)=2.5, FPS=196, Pre-training=scratch2022.09 | 79.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| StairNet + SEBackbone=VGG16, Detector=StairNet + SE, Parameters (M)=32.12018.07 | 79.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| BlitzNet300backbone=ResNet502018.09 | 79.1 | 86.7 | 86.2 | 78.9 | 73.1 | 47.6 | 85.7 | 86.1 | 87.7 | 59.3 | 85.1 | 78.4 | 86.3 | 87.9 | 87.8 | 76.8 | 51.8 | 78.4 | 81.3 | 81.7 | 85.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FCOSBackbone=Conv R50, Neck=Conv2021.12 | 79.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| StairNetBackbone=VGG16, Detector=StairNet, Parameters (M)=32.02018.07 | 78.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| OHEMBackbone=VGG16, Multi-scale (M)=true, Multi-stage bbox regression (B)=true, Training Set=VOC07+12 trainval2016.04 | 78.9 | 80.6 | 85.7 | 79.8 | 69.9 | 60.8 | 88.3 | 87.9 | 89.6 | 59.7 | 85.1 | 76.5 | 87.1 | 87.3 | 82.4 | 78.8 | 53.7 | 80.5 | 78.7 | 84.5 | 80.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RPN+VGGdetector=Fast R-CNN, backbone=VGG-16, feature_sharing=shared, num_proposals=300, training_data=COCO + VOC 2007 trainval + VOC 2012 trainval2015.06 | 78.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Faster R-CNNTraining Data=COCO trainval35k then VOC07+12, Input Size=min dim: 6002015.12 | 78.8 | 84.3 | 82 | 77.7 | 68.9 | 65.7 | 88.1 | 88.4 | 88.9 | 63.6 | 86.3 | 70.8 | 85.9 | 87.6 | 80.1 | 82.3 | 53.6 | 80.4 | 75.8 | 86.6 | 78.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Faster R-CNNTraining Data=COCO + 07+122017.07 | 78.8 | 84.3 | 82 | 77.7 | 68.9 | 65.7 | 88.1 | 88.4 | 88.9 | 63.6 | 86.3 | 70.8 | 85.9 | 87.6 | 80.1 | 82.3 | 53.6 | 80.4 | 75.8 | 86.6 | 78.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Faster R-CNNBackbone=VGG-16, Stage=two-stage, Multi-scale testing=false2017.11 | 78.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| AdamBackbone=ResNet50, Framework=Faster-RCNN, Neck=FPN, Pre-training=COCO dataset2026.03 | 78.67 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| YOLOv2Backbone=Darknet-19, Input size=544 x 544, #Boxes=845, FPS=402017.11 | 78.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DSSD321Backbone=ResNet-101, Input size=321 x 321, #Boxes=17080, FPS=9.52017.11 | 78.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DSSD321backbone=ResNet1012018.09 | 78.6 | 81.9 | 84.9 | 80.5 | 68.4 | 53.9 | 85.6 | 86.2 | 88.9 | 61.1 | 83.5 | 78.7 | 86.7 | 88.7 | 86.9 | 79.7 | 51.7 | 78 | 80.9 | 87.2 | 79.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ThunderNetYear=ICCV2019, Backbone=SNet-535, Input Size=320 x 320, FLOPS (G)=1.3, FPS=2142022.09 | 78.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| AdamWBackbone=ResNet50, Framework=Faster-RCNN, Neck=FPN, Pre-training=COCO dataset2026.03 | 78.48 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MR-CNNBackbone=VGG16, Multi-scale (M)=true, Multi-stage bbox regression (B)=true, Training Set=VOC07+12 trainval2016.04 | 78.2 | 80.3 | 84.1 | 78.5 | 70.8 | 68.5 | 88 | 85.9 | 87.8 | 60.3 | 85.2 | 73.7 | 87.2 | 86.5 | 85 | 76.4 | 48.5 | 76.3 | 75.5 | 85 | 81 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MR-CNNBackbone=VGG-16, Input size=~1000 × 600, #Boxes=250, FPS=0.032017.11 | 78.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| STDN300backbone=DenseNet1692018.09 | 78.1 | 81.1 | 86.9 | 76.4 | 69.2 | 52.4 | 87.7 | 84.2 | 88.3 | 60.2 | 81.3 | 77.6 | 86.6 | 88.9 | 84.2 | 76.8 | 51.8 | 78.4 | 81.3 | 87.5 | 77.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SSDBackbone=VGG16, Detector=SSD, Parameters (M)=26.52018.07 | 77.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DSOD300Backbone=DS/64-192-48-1, Input size=300 x 300, #Boxes=8732, FPS=17.42017.11 | 77.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SSDBackbone=ResNet-50, Augmentation=CutMix2019.05 | 77.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SSD300*Train=07+12, Base Network=VGG162017.08 | 77.5 | 79.5 | 83.9 | 76 | 69.6 | 50.5 | 87 | 85.7 | 88.1 | 60.3 | 81.5 | 77 | 86.1 | 87.5 | 83.9 | 79.4 | 52.3 | 77.9 | 79.5 | 87.6 | 76.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SSD300Implementations=Original2017.08 | 77.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SSD300Implementations=ChainerCV2017.08 | 77.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SSD300backbone=VGG162018.09 | 77.5 | 79.5 | 83.9 | 76 | 69.6 | 50.5 | 87 | 85.7 | 88.1 | 60.3 | 81.5 | 77 | 86.1 | 87.5 | 84 | 79.4 | 52.3 | 77.9 | 79.5 | 87.6 | 76.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Teacher (T2)Network=VGG-SSD, # of params=26.3M2019.04 | 77.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RetinaNetBackbone=Conv R50, Neck=Conv2021.12 | 77.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SSD300*Backbone=VGG-16, Input size=300 x 300, #Boxes=8732, FPS=462017.11 | 77.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DiCENetImage size=512x512, FLOPS=2.0 B2019.06 | 77.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VGG + SSDParameters=26.2M, FLOPs=31B2021.07 | 77.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SSD321Backbone=ResNet-101, Input size=321 x 321, #Boxes=17080, FPS=11.22017.11 | 77.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DFPR300backbone=ResNet1012018.09 | 77.1 | 89.3 | 84.9 | 79.9 | 75.6 | 55.4 | 88.2 | 88.6 | 88.6 | 63.3 | 87.9 | 78.8 | 87.3 | 87.7 | 85.5 | 80.5 | 55.4 | 81.1 | 79.6 | 87.8 | 78.5 | — | — | — | — | — | — | — | — | — | — | — | — | — |