Research map / Detection and segmentation / CNN features
Deep Learning Markov Random Field for Semantic Segmentation
Ziwei Liu, Xiaoxiao Li, Ping Luo, Chen Change Loy, Xiaoou Tang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add triple penalty (b12) to unary term (ablation) | DPN · VOC12 val | mIoU: 62.4 -> 64.7 | Improved |
| add mixture of label contexts (b13) (ablation) | DPN · VOC12 val | mIoU: 64.7 -> 66.5 | Improved |
| add joint fine-tuning of all components (ablation) | DPN · VOC12 val | mIoU: 66.5 -> 67.8 | Improved |
| enlarge b12 receptive field from 10x10 to 50x50 (ablation) | DPN · VOC12 val | mIoU: 63.8 -> 64.7 | Improved |
| enlarge b12 receptive field from 50x50 to 100x100 (ablation) | DPN · VOC12 val | mIoU: 64.7 -> 64.3 | No clear effect |
| enlarge b13 receptive field from 1x1 to 5x5 (ablation) | DPN · VOC12 val | mIoU: 64.8 -> 66.0 | Improved |
| use 9x9 mixtures instead of single 9x9 filter in b13 (ablation) | DPN · VOC12 val | mIoU: 66.3 -> 66.5 | Improved |
| use DPN pairwise terms instead of DSN or DeepLab pairwise terms (ablation) | DPN · VOC12 val | improvement over unary (%): 2.6 -> 5.4 | Improved |
| use 3D spatial-temporal pairwise term instead of 2D | DPN · CamVid test | mIoU: 60.06 -> 60.25 | Improved |
| incremental learning instead of joint learning from the start (ablation) | DPN · VOC12 val | training accuracy curve | Improved |
| use one MF iteration instead of 5 or more (ablation) | DPN · VOC12 val | accuracy | No clear effect |
| add COCO pre-training | DPN · VOC12 test | mIoU: 74.1 -> 77.5 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SegNet — SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation (TPAMI 2017) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Efficient Piecewise Training of Deep Structured Models for Semantic Segmentation (CVPR 2016 · similarity 0.89)
- Gaussian Conditional Random Field Network for Semantic Segmentation (CVPR 2016 · similarity 0.88)
- Exploring Context with Deep Structured Models for Semantic Segmentation (TPAMI 2017 · similarity 0.88)
- CNN in MRF: Video Object Segmentation via Inference in a CNN-Based Higher-Order Spatio-Temporal MRF (CVPR 2018 · similarity 0.88)
- Semantic Segmentation With Boundary Neural Fields (CVPR 2016 · similarity 0.87)
- MRFP: Learning Generalizable Semantic Segmentation from Sim-2-Real with Multi-Resolution Feature Perturbation (CVPR 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
