| Rethinking Pre-training and Self-training | ✓ Link | 90.0% | | | EfficientNet-L2+NAS-FPN (single scale test, with self-training) | 2020-06-11 |
| Text-image Alignment for Diffusion-based Perception | ✓ Link | 87.11% | | | TADP | 2023-09-29 |
| Simple Copy-Paste is a Strong Data Augmentation Method for Instance Segmentation | ✓ Link | 86.6% | | | Eff-B7 NAS-FPN (Copy-Paste pre-training, single-scale)) | 2020-12-13 |
| ExFuse: Enhancing Feature Fusion for Semantic Segmentation | | 85.8% | | | ExFuse (ResNeXt-131) | 2018-04-11 |
| Dilated SpineNet for Semantic Segmentation | | 85.64% | | | SpineNet-S143 (single-scale test) | 2021-03-23 |
| Rethinking Atrous Convolution for Semantic Image Segmentation | ✓ Link | 82.7% | | | DeepLabv3-JFT | 2017-06-17 |
| Auto-DeepLab: Hierarchical Neural Architecture Search for Semantic Image Segmentation | ✓ Link | 82.04% | | | Auto-DeepLab-L | 2019-01-10 |
| Large Kernel Matters -- Improve Semantic Segmentation by Global Convolutional Network | ✓ Link | 81.0% | | | ResNet-GCN | 2017-03-08 |
| HyperSeg: Patch-wise Hypernetwork for Real-time Semantic Segmentation | ✓ Link | 80.61% | | | HyperSeg-L | 2020-12-21 |
| Learning a Discriminative Feature Network for Semantic Segmentation | ✓ Link | 80.60% | | | DFN (ResNet-101) | 2018-04-25 |
| Waterfall Atrous Spatial Pooling Architecture for Efficient Semantic Segmentation | ✓ Link | 80.41% | | | WASPnet-CRF (ours) | 2019-12-06 |
| Res2Net: A New Multi-scale Backbone Architecture | ✓ Link | 79.3% | | | Deeplab v3+ (Res2Net-101) | 2019-04-02 |
| Fast Neural Architecture Search of Compact Semantic Segmentation Models via Auxiliary Cells | ✓ Link | 78.0% | | | FastDenseNas-arch0 | 2018-10-25 |
| Pushing the limits of self-supervised ResNets: Can we outperform supervised learning without labels on ImageNet? | ✓ Link | 77.9% | | | ReLICv2 | 2022-01-13 |
| DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs | ✓ Link | 77.69% | | | DeepLab-CRF (ResNet-101) | 2016-06-02 |
| Fast Neural Architecture Search of Compact Semantic Segmentation Models via Auxiliary Cells | ✓ Link | 77.3% | | | FastDenseNas-arch2 | 2018-10-25 |
| Pushing the limits of self-supervised ResNets: Can we outperform supervised learning without labels on ImageNet? | ✓ Link | 77.3% | | | DetCon | 2022-01-13 |
| Fast Neural Architecture Search of Compact Semantic Segmentation Models via Auxiliary Cells | ✓ Link | 77.1% | | | FastDenseNas-arch1 | 2018-10-25 |
| Revisiting Unreasonable Effectiveness of Data in Deep Learning Era | ✓ Link | 76.5% | | | DeepLabv3 (ImageNet+300M) | 2017-07-10 |
| Pushing the limits of self-supervised ResNets: Can we outperform supervised learning without labels on ImageNet? | ✓ Link | 75.7% | | | BYOL | 2022-01-13 |
| DiCENet: Dimension-wise Convolutions for Efficient Networks | ✓ Link | 66.5% | | | DiCENet | 2019-06-08 |
| Reliability Does Matter: An End-to-End Weakly Supervised Semantic Segmentation Approach | ✓ Link | 66.3 | | | RRM | 2019-11-19 |
| Scaling up Multi-domain Semantic Segmentation with Sentence Embeddings | | 65% | | | SIW | 2022-02-04 |
| Self-Supervised Difference Detection for Weakly-Supervised Semantic Segmentation | ✓ Link | 64.9 | | | SSDD | 2019-11-04 |
| Find it if You Can: End-to-End Adversarial Erasing for Weakly-Supervised Semantic Segmentation | ✓ Link | 62.8% | | | PSA w/ EADER DeepLab (Xception-65) | 2020-11-09 |
| Exploiting saliency for object segmentation from image level labels | | 55.7% | | | G2 | 2017-01-28 |
| Weakly Supervised Instance Segmentation using Class Peak Response | ✓ Link | 53.4% | | | PRM | 2018-04-03 |
| Simple Does It: Weakly Supervised Instance and Semantic Segmentation | | | 71.6 | | SID | 2016-03-24 |
| Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation | ✓ Link | | | 75.39 | DeepLabV3+ (ResNet-101) | 2018-02-07 |