Semantic Part Segmentation on PASCAL-Person-Part (val)
68.6mIoURefineNet
Evaluation results
| Method | Links | |
|---|---|---|
| RefineNet2017.07 | 68.6 | |
| RAN-sstructure=MSC, decision filters=3-by-3, dilation=false2017.07 | 66.6 | |
| RAN-nstructure=MSC, decision filters=3-by-3, dilation=false2017.07 | 66.5 | |
| DeepLabBackbone=ResNet-101, Multi-scale=true, MS-COCO pre-training=true, Data augmentation=true, CRF=true2016.06 | 64.94 | |
| DeepLabv22017.07 | 64.9 | |
| DeepLabBackbone=ResNet-101, Multi-scale=true, MS-COCO pre-training=true, Data augmentation=true2016.06 | 64.4 | |
| DeepLabBackbone=ResNet-101, Multi-scale=true2016.06 | 63.1 | |
| DeepLabBackbone=ResNet-101, Multi-scale=true, MS-COCO pre-training=true, Data augmentation=true, LargeFOV=true2016.06 | 62.76 | |
| DeepLabBackbone=ResNet-101, Multi-scale=true, MS-COCO pre-training=true, Data augmentation=true, ASPP=true2016.06 | 62.18 | |
| Graph LSTM2017.07 | 60.2 | |
| Graph LSTM2016.06 | 60.16 | |
| DeepLabBackbone=ResNet-1012016.06 | 58.9 | |
| LG-LSTM2016.06 | 57.97 | |
| HAZN2016.06 | 57.54 | |
| HAZN2017.07 | 57.5 | |
| Attention2017.07 | 56.4 | |
| AttentionBackbone=VGG-162016.06 | 56.39 |