Research map / Detection and segmentation / CNN features
Generating Dynamic Kernels via Transformers for Lane Detection
Ziye Chen, Yu Liu, Mingming Gong, Bo Du, Guoqi Qian, Kate Smith-Miles
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 |
|---|---|---|---|
| increase number of lane queries from 20 to 80 (ablation) | transformer dynamic kernel lane detector · CurveLanes | F1: 87.37 -> 87.99 | Improved |
| increase number of lane queries from 80 to 100 (ablation) | transformer dynamic kernel lane detector · CurveLanes | F1: 87.99 -> 87.97 | No clear effect |
| increase number of lane queries from 20 to 80 (ablation) | transformer dynamic kernel lane detector · CULane | F1: 80.13 -> 80.36 | Improved |
| increase decoder layers from 1 to 4 (ablation) | transformer dynamic kernel lane detector · CULane | F1: 79.48 -> 80.36 | Improved |
| increase decoder layers from 1 to 4 (ablation) | transformer dynamic kernel lane detector · CurveLanes | F1: 86.82 -> 87.99 | Improved |
| increase decoder layers from 4 to 6 (ablation) | transformer dynamic kernel lane detector · CULane | F1: 80.36 -> 80.44 | No clear effect |
| generate dynamic kernels with a transformer instead of from lane key points (CondLaneNet) | dynamic-convolution lane detector · OpenLane | F1: 59.1 -> 63.4 (ResNet-101) | Improved |
| generate dynamic kernels with a transformer instead of from lane key points (CondLaneNet) | dynamic-convolution lane detector · CurveLanes | F1: 86.10 -> 88.47 (ResNet-101) | Improved |
| generate dynamic kernels with a transformer instead of from lane key points (CondLaneNet) | dynamic-convolution lane detector · TuSimple | F1: 97.24 -> 97.94 (ResNet-101) | No clear effect |
| transformer kernel generation costs more compute than CondLaneNet | ResNet-18 model · OpenLane | FPS / GFlops: 173 -> 105 FPS; 10.2 -> 13.7 GFlops | Hurt |
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.
- PersFormer — PersFormer: 3D Lane Detection via Perspective Transformer and the OpenLane Benchmark (ECCV 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 0.90)
- Laneformer: Object-Aware Row-Column Transformers for Lane Detection (AAAI 2022 · similarity 0.90)
- A Siamese Transformer with Hierarchical Refinement for Lane Detection (NeurIPS 2024 · similarity 0.88)
- GEONet: Global Enhancement and Optimization Network for Lane Detection (AAAI 2025 · similarity 0.88)
- LaneCPP: Continuous 3D Lane Detection using Physical Priors (CVPR 2024 · similarity 0.87)
- LaneSegNet: Map Learning with Lane Segment Perception for Autonomous Driving (ICLR 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.
