Research map / Detection and segmentation / CNN features
SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation
Vijay Badrinarayanan, Alex Kendall, Roberto Cipolla
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 |
|---|---|---|---|
| learn the upsampling instead of fixed bilinear interpolation (ablation) | SegNet-Basic decoder · CamVid test | mean I/U: 45.6 -> 46.3 | Improved |
| use max-pooling indices (SegNet-Basic) instead of FCN-style decoder (ablation) | SegNet-Basic vs FCN-Basic · CamVid test | mean I/U: 46.4 -> 46.3 | No clear effect |
| add encoder feature map addition to FCN decoder (ablation) | FCN-Basic · CamVid test | mean I/U: 42.7 -> 46.4 | Improved |
| add encoder feature map addition to SegNet decoder (ablation) | SegNet-Basic · CamVid test | mean I/U: 46.3 -> 48.0 | Improved |
| use multi-channel decoder filters instead of single-channel (ablation) | SegNet-Basic · CamVid test | mean I/U: 44.6 -> 46.3 | Improved |
| remove dimensionality reduction of encoder maps in FCN (ablation) | FCN-Basic · CamVid test | mean I/U: 46.4 -> 47.6 | Improved |
| use median frequency class balancing instead of natural frequency (ablation) | SegNet-Basic · CamVid test | mean I/U: 38.0 -> 46.3 | Improved |
| use median frequency class balancing instead of natural frequency (ablation) | SegNet-Basic · CamVid test | global accuracy: 84.1 -> 82.8 | Hurt |
| use pretrained VGG16 encoder and deeper SegNet instead of SegNet-Basic | SegNet · CamVid test | mean I/U: 46.3 -> 50.2 | Improved |
| train on larger 3.5K road scene dataset | SegNet · CamVid test | class average accuracy: 65.9 -> 81.3 | Improved |
| train on larger 3.5K road scene dataset | SegNet · CamVid test | mean I/U: 50.2 -> 69.1 | Improved |
| train encoder-decoder layer-wise with L-BFGS instead of SGD | SegNet-Basic · CamVid test | class average accuracy: 62.3 -> 62.9 | Improved |
Later papers that compared with this method
15 later accepted papers report SegNet as a baseline in their results tables; 14 of them report a better number on most of the results they share with it.
- CrackFormer: Transformer Network for Fine-Grained Crack Detection (ICCV 2021 · better on 3 of 3 shared results)
- Severity-Aware Semantic Segmentation With Reinforced Wasserstein Training (CVPR 2020 · better on 2 of 2 shared results)
- GMNet: Graph Matching Network for Large Scale Part Semantic Segmentation in the Wild (ECCV 2020 · better on 2 of 2 shared results)
- Importance-Aware Semantic Segmentation in Self-Driving with Discrete Wasserstein Training (AAAI 2020 · better on 1 of 1 shared result)
- Multi-Class Part Parsing With Joint Boundary-Semantic Awareness (ICCV 2019 · better on 2 of 2 shared results)
- HarDNet: A Low Memory Traffic Network (ICCV 2019 · better on 1 of 1 shared result)
- Geometric Constrained Joint Lane Segmentation and Lane Boundary Detection (ECCV 2018 · better on 6 of 6 shared results)
- SketchyScene: Richly-Annotated Scene Sketches (ECCV 2018 · better on 6 of 6 shared results)
7 more not listed.
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
- SegNeXt: Rethinking Convolutional Attention Design for Semantic Segmentation (NeurIPS 2022 · similarity 0.92)
- DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs (TPAMI 2017 · similarity 0.89)
- SPGNet: Semantic Prediction Guidance for Scene Parsing (ICCV 2019 · similarity 0.89)
- SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers (NeurIPS 2021 · similarity 0.88)
- EfficientFCN: Holistically-guided Decoding for Semantic Segmentation (ECCV 2020 · similarity 0.88)
- SegMAN: Omni-scale Context Modeling with State Space Models and Local Attention for Semantic Segmentation (CVPR 2025 · similarity 0.87)
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.
