Boundary Detection on BSDS 500 (test)
84.8ODSEDTER
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| EDTERBackbone=Transformer, Pub.'Year=CVPR'22, Testing Scale=Multi-scale (MS), Extra Training Data=PASCAL VOC2023.03 | 84.8 | 86.5 | 90.3 | — | |
| UAEDBackbone=EfficientNet, Testing Scale=Multi-scale (MS), Extra Training Data=PASCAL VOC2023.03 | 84.4 | 86.4 | 90.5 | — | |
| EDTERBackbone=Transformer, Pub.'Year=CVPR'22, Testing Scale=Multi-scale (MS), Extra Training Data=None2023.03 | 84 | 85.8 | 89.6 | — | |
| UAEDBackbone=EfficientNet, Testing Scale=Single-scale (SS), Extra Training Data=PASCAL VOC2023.03 | 83.8 | 85.5 | 90.2 | — | |
| UAEDBackbone=EfficientNet, Testing Scale=Multi-scale (MS), Extra Training Data=None2023.03 | 83.7 | 85.5 | 89.7 | — | |
| EDTERBackbone=Transformer, Pub.'Year=CVPR'22, Testing Scale=Single-scale (SS), Extra Training Data=PASCAL VOC2023.03 | 83.2 | 84.7 | 88.6 | — | |
| UAEDBackbone=VGG16, Testing Scale=Multi-scale (MS), Extra Training Data=PASCAL VOC2023.03 | 83 | 85 | 89.5 | — | |
| UAEDBackbone=EfficientNet, Testing Scale=Single-scale (SS), Extra Training Data=None2023.03 | 82.9 | 84.7 | 89.2 | — | |
| BDCNmode=multi-scale fusion2019.02 | 82.8 | 84.4 | 0.89 | — | |
| BDCNBackbone=VGG16, Pub.'Year=CVPR'19, Testing Scale=Multi-scale (MS), Extra Training Data=PASCAL VOC2023.03 | 82.8 | 84.4 | 89 | — | |
| FCL-NetBackbone=VGG16, Pub.'Year=NN'22, Testing Scale=Multi-scale (MS), Extra Training Data=PASCAL VOC2023.03 | 82.6 | 84.5 | — | — | |
| EDTERBackbone=Transformer, Pub.'Year=CVPR'22, Testing Scale=Single-scale (SS), Extra Training Data=None2023.03 | 82.4 | 84.1 | 88 | — | |
| BDCNadditional training data=PASCAL-Context data2019.02 | 82 | 83.8 | 0.888 | — | |
| BDCNBackbone=VGG16, Pub.'Year=CVPR'19, Testing Scale=Single-scale (SS), Extra Training Data=PASCAL VOC2023.03 | 82 | 83.8 | 88.8 | — | |
| UAEDBackbone=VGG16, Testing Scale=Single-scale (SS), Extra Training Data=PASCAL VOC2023.03 | 82 | 84 | 88.9 | — | |
| BDCNFPS=472021.08 | 82 | 83.8 | — | — | |
| BDCNParams=16.3M, MACs=23.2G, Throughput=47FPS, Pre-training=ImageNet2021.08 | 82 | — | — | — | |
| LDCBackbone=MobileNetV2, Pub.'Year=ACMMM'21, Testing Scale=Multi-scale (MS), Extra Training Data=PASCAL VOC2023.03 | 81.9 | 83.4 | 86 | — | |
| UAEDBackbone=VGG16, Testing Scale=Multi-scale (MS), Extra Training Data=None2023.03 | 81.9 | 83.8 | 88.1 | — | |
| FCL-NetBackbone=VGG16, Pub.'Year=NN'22, Testing Scale=Multi-scale (MS), Extra Training Data=None2023.03 | 81.6 | 83.3 | — | — | |
| CEDmode=multi-scale fusion2019.02 | 81.5 | 83.3 | 0.889 | — | |
| LPCBmode=multi-scale fusion2019.02 | 81.5 | 83.4 | — | — | |
| CEDBackbone=VGG16, Pub.'Year=CVPR'17, Testing Scale=Single-scale (SS), Extra Training Data=PASCAL VOC2023.03 | 81.5 | 83.3 | 88.9 | — | |
| LPCBBackbone=VGG16, Pub.'Year=ECCV'18, Testing Scale=Multi-scale (MS), Extra Training Data=PASCAL VOC2023.03 | 81.5 | 83.4 | — | — | |
| FCL-NetBackbone=VGG16, Pub.'Year=NN'22, Testing Scale=Single-scale (SS), Extra Training Data=PASCAL VOC2023.03 | 81.5 | 83.4 | — | — | |
| Deep Boundarymode=multi-scale fusion, post-processing=Grouping2019.02 | 81.3 | 83.1 | 0.866 | — | |
| Deep BoundaryBackbone=VGG16, Pub.'Year=ICLR'15, Testing Scale=Multi-scale (MS), Extra Training Data=PASCAL VOC2023.03 | 81.3 | 83.1 | 86.6 | — | |
| CATS-BDCNEvaluation protocol=Standard (SEval), Scale=Single-scale, Training data=BSDS500 only, Backbone=VGG162020.11 | 81.2 | 82.8 | — | — | |
| LDCBackbone=MobileNetV2, Pub.'Year=ACMMM'21, Testing Scale=Single-scale (SS), Extra Training Data=PASCAL VOC2023.03 | 81.2 | 82.6 | 85.7 | — | |
| RCFmode=multi-scale fusion2019.02 | 81.1 | 83 | — | — | |
| RCFBackbone=VGG16, Pub.'Year=CVPR'17, Testing Scale=Multi-scale (MS), Extra Training Data=PASCAL VOC2023.03 | 81.1 | 83 | 84.6 | — | |
| Deep Boundarymode=multi-scale fusion2019.02 | 80.9 | 82.7 | 0.861 | — | |
| Deep BoundaryBackbone=VGG16, Pub.'Year=ICLR'15, Testing Scale=Single-scale (SS), Extra Training Data=PASCAL VOC2023.03 | 80.9 | 82.7 | 86.1 | — | |
| LPCBadditional training data=PASCAL-Context data2019.02 | 80.8 | 82.4 | — | — | |
| LPCBBackbone=VGG16, Pub.'Year=ECCV'18, Testing Scale=Single-scale (SS), Extra Training Data=PASCAL VOC2023.03 | 80.8 | 82.4 | — | — | |
| UAEDBackbone=VGG16, Testing Scale=Single-scale (SS), Extra Training Data=None2023.03 | 80.8 | 82.7 | 87.2 | — | |
| LPCBFPS=30+2021.08 | 80.8 | 82.4 | — | — | |
| BDCNEvaluation protocol=Standard (SEval), Scale=Single-scale, Training data=BSDS500 only, Backbone=VGG162020.11 | 80.7 | 82.2 | — | — | |
| PiDiNetBackbone=PDC, Pub.'Year=ICCV'21, Testing Scale=Single-scale (SS), Extra Training Data=None2023.03 | 80.7 | 82.2 | — | — | |
| PiDiNetFPS=92‡,*2021.08 | 80.7 | 82.3 | — | — | |
| PiDiNetParams=710K, MACs=3.43G, Throughput=92FPS, Pre-training=No2021.08 | 80.7 | — | — | — | |
| RCF2019.09 | 80.6 | 82.3 | — | — | |
| RCFadditional training data=PASCAL-Context data2019.02 | 80.6 | 82.3 | — | — | |
| BDCN2019.02 | 80.6 | 82.6 | 0.847 | — | |
| RCFBackbone=VGG16, Pub.'Year=CVPR'17, Testing Scale=Single-scale (SS), Extra Training Data=PASCAL VOC2023.03 | 80.6 | 82.3 | — | — | |
| BDCNBackbone=VGG16, Pub.'Year=CVPR'19, Testing Scale=Single-scale (SS), Extra Training Data=None2023.03 | 80.6 | 82.6 | 84.7 | — | |
| RCFFPS=672021.08 | 80.6 | 82.3 | — | — | |
| RCFParams=14.8M, MACs=16.2G, Throughput=67FPS, Pre-training=ImageNet2021.08 | 80.6 | — | — | — | |
| CATS-RCFEvaluation protocol=Standard (SEval), Scale=Single-scale, Training data=BSDS500 only, Backbone=VGG162020.11 | 80.5 | 82.2 | — | — | |
| CED2019.09 | 80.3 | 82 | 87.1 | — | |
| Human2019.02 | 80.3 | 80.3 | — | — | |
| HumanEvaluation protocol=Standard (SEval), Scale=Single-scale, Training data=BSDS500 only2020.11 | 80.3 | 80.3 | — | — | |
| Deep BoundaryBackbone=VGG16, Pub.'Year=ICLR'15, Testing Scale=Multi-scale (MS), Extra Training Data=None2023.03 | 80.3 | 82 | 84.8 | — | |
| CEDBackbone=VGG16, Pub.'Year=CVPR'17, Testing Scale=Single-scale (SS), Extra Training Data=None2023.03 | 80.3 | 82 | 87.1 | — | |
| Human2021.08 | 80.3 | 80.3 | — | — | |
| DSCDBackbone=VGG16, Pub.'Year=ACMMM'20, Testing Scale=Single-scale (SS), Extra Training Data=None2023.03 | 80.2 | 81.7 | — | — | |
| LPCB2019.02 | 80 | 81.6 | — | — | |
| LPCBEvaluation protocol=Standard (SEval), Scale=Single-scale, Training data=BSDS500 only2020.11 | 80 | 81.6 | — | — | |
| CATS-HEDEvaluation protocol=Standard (SEval), Scale=Single-scale, Training data=BSDS500 only, Backbone=VGG162020.11 | 80 | 81.6 | — | — | |
| LPCBBackbone=VGG16, Pub.'Year=ECCV'18, Testing Scale=Single-scale (SS), Extra Training Data=None2023.03 | 80 | 81.6 | — | — | |
| PiDiNet-LFPS=128+2021.08 | 80 | 81.5 | — | — | |
| COB2019.02 | 79.9 | 81.7 | 0.849 | — | |
| DCDEvaluation protocol=Standard (SEval), Scale=Single-scale, Training data=BSDS500 only2020.11 | 79.9 | 81.7 | — | — | |
| RCFEvaluation protocol=Standard (SEval), Scale=Single-scale, Training data=BSDS500 only, Backbone=VGG162020.11 | 79.9 | 81.5 | — | — | |
| LDCBackbone=MobileNetV2, Pub.'Year=ACMMM'21, Testing Scale=Single-scale (SS), Extra Training Data=None2023.03 | 79.9 | 81.6 | 83.7 | — | |
| DCD2019.02 | 79.8 | 82.9 | 0.869 | — | |
| AMH-Net2019.02 | 79.8 | 81.5 | — | — | |
| AMH-NetEvaluation protocol=Standard (SEval), Scale=Single-scale, Training data=BSDS500 only2020.11 | 79.8 | 82.9 | — | — | |
| AMH-NetBackbone=ResNet50, Pub.'Year=NeurIPS'17, Testing Scale=Single-scale (SS), Extra Training Data=None2023.03 | 79.8 | 82.9 | 86.9 | — | |
| RCFBackbone=VGG16, Pub.'Year=CVPR'17, Testing Scale=Single-scale (SS), Extra Training Data=None2023.03 | 79.8 | 81.5 | — | — | |
| AMH-Net2021.08 | 79.8 | 82.9 | — | — | |
| BaselineFPS=96+2021.08 | 79.8 | 81.6 | — | — | |
| PiDiNet-SmallFPS=148+2021.08 | 79.8 | 81.4 | — | — | |
| CED2019.02 | 79.4 | 81.1 | 0.847 | — | |
| CEDEvaluation protocol=Standard (SEval), Scale=Single-scale, Training data=BSDS500 only2020.11 | 79.4 | 81.1 | — | — | |
| CED2021.08 | 79.4 | 81.1 | — | — | |
| RCF2019.02 | 79.3 | 82 | 0.859 | — | |
| COBEvaluation protocol=Standard (SEval), Scale=Single-scale, Training data=BSDS500 only2020.11 | 79.3 | 82 | — | — | |
| HEDEvaluation protocol=Standard (SEval), Scale=Single-scale, Training data=BSDS500 only, Backbone=VGG162020.11 | 79.3 | 81.1 | — | — | |
| COBBackbone=VGG16, Pub.'Year=ECCV'16, Testing Scale=Single-scale (SS), Extra Training Data=None2023.03 | 79.3 | 82 | 85.9 | — | |
| COB2021.08 | 79.3 | 82 | — | — | |
| PiDiNet-Small-LFPS=212+2021.08 | 79.3 | 80.9 | — | — | |
| RDSBackbone=VGG16, Pub.'Year=CVPR'16, Testing Scale=Single-scale (SS), Extra Training Data=None2023.03 | 79.2 | 81 | 81.8 | — | |
| HED(BSDS) cons.Family=Trained on ground truth, Training Data=BSDS Ground Truth (consensus)2015.11 | 79 | 81 | 84 | — | |
| HED2019.09 | 79 | 80.8 | 81.1 | — | |
| FINED-TrainFPS=99+2021.08 | 79 | 80.8 | — | — | |
| Deep Boundary2019.02 | 78.9 | 81.1 | 0.789 | — | |
| DeepBoundaryEvaluation protocol=Standard (SEval), Scale=Single-scale, Training data=BSDS500 only2020.11 | 78.9 | 81.1 | — | — | |
| Deep BoundaryBackbone=VGG16, Pub.'Year=ICLR'15, Testing Scale=Single-scale (SS), Extra Training Data=None2023.03 | 78.9 | 81.1 | 78.9 | — | |
| DeepBoundary2021.08 | 78.9 | 81.1 | — | — | |
| PiDiNet-TinyFPS=152+2021.08 | 78.9 | 80.6 | — | — | |
| HED2019.02 | 78.8 | 80.8 | 0.84 | — | |
| CEDNadditional training data=PASCAL-Context data2019.02 | 78.8 | 80.4 | — | — | |
| CEDNEvaluation protocol=Standard (SEval), Scale=Single-scale, Training data=BSDS500 only2020.11 | 78.8 | 80.4 | — | — | |
| HEDBackbone=VGG16, Pub.'Year=ICCV'15, Testing Scale=Single-scale (SS), Extra Training Data=None2023.03 | 78.8 | 80.8 | 84 | — | |
| CEDNBackbone=VGG16, Pub.'Year=CVPR'16, Testing Scale=Single-scale (SS), Extra Training Data=None2023.03 | 78.8 | 80.4 | — | — | |
| CEDNFPS=10+2021.08 | 78.8 | 80.4 | — | — | |
| HEDFPS=78+2021.08 | 78.8 | 80.8 | — | — | |
| FINED-InfFPS=124+2021.08 | 78.8 | 80.4 | — | — | |
| HEDParams=14.7M, MACs=22.2G, Throughput=78FPS, Pre-training=ImageNet2021.08 | 78.8 | — | — | — |