Research map / Detection and segmentation / CNN features
RESA: Recurrent Feature-Shift Aggregator for Lane Detection
Tu Zheng, Hao Fang, Yi Zhang, Wenjian Tang, Zheng Yang, Haifeng Liu, Deng Cai
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 RESA to a ResNet-34 baseline (ablation) | ResNet-34 lane detector · CULane | F1: 65.1 -> 74.3 | Improved |
| Replace direct bilinear up-sampling with BUSD (ablation) | ResNet-34 lane detector · CULane | F1: 65.1 -> 68.6 | Improved |
| Add both RESA and BUSD (ablation) | ResNet-34 lane detector · CULane | F1: 65.1 -> 74.5 | Improved |
| Add spatial aggregation directions progressively (ablation) | RESA with ResNet-34 · CULane | F1-measure: 72.4 -> 73.7 -> 74.4 -> 74.5 | Improved |
| Replace addition with maximum feature aggregation (ablation) | RESA_DULR · CULane | F1-measure: 74.5 -> 74.4 | No clear effect |
| Increase RESA iterations from 1 to 4 (ablation) | RESA with ResNet-34 · CULane | F1-measure: 73.2 -> 74.5 | Improved |
| Increase RESA iterations from 4 to 5 (ablation) | RESA with ResNet-34 · CULane | F1-measure: 74.5 -> 74.5 | No clear effect |
| Use RESA instead of SCNN (ablation) | VGG16 lane detector · CULane | F1-measure: 72.3 -> 73.3 | Improved |
| Use RESA instead of SCNN (ablation) | ResNet-34 lane detector · CULane | F1-measure: 72.7 -> 74.5 | Improved |
| Use RESA rather than SCNN message passing | RESA and SCNN runtime comparison, kernel width 7 · Runtime comparison | runtime (ms): 43.5 -> 3.8 | Improved |
| Use RESA rather than SCNN message passing | RESA and SCNN runtime comparison, kernel width 9 · Runtime comparison | runtime (ms): 44.0 -> 4.0 | Improved |
| Use RESA rather than SCNN message passing | RESA and SCNN runtime comparison, kernel width 11 · Runtime comparison | runtime (ms): 44.6 -> 4.4 | Improved |
Later papers that compared with this method
1 later accepted paper reports RESA as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Ultra Fast Deep Lane Detection With Hybrid Anchor Driven Ordinal Classification (TPAMI 2022 · 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
- Sketch and Refine: Towards Fast and Accurate Lane Detection (AAAI 2024 · similarity 0.91)
- Laneformer: Object-Aware Row-Column Transformers for Lane Detection (AAAI 2022 · similarity 0.89)
- FastDraw: Addressing the Long Tail of Lane Detection by Adapting a Sequential Prediction Network (CVPR 2019 · similarity 0.89)
- GEONet: Global Enhancement and Optimization Network for Lane Detection (AAAI 2025 · similarity 0.89)
- A Siamese Transformer with Hierarchical Refinement for Lane Detection (NeurIPS 2024 · similarity 0.89)
- GSENet:Global Semantic Enhancement Network for Lane Detection (AAAI 2024 · 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.
