Research map / Detection and segmentation / CNN features
Towards Human-Level License Plate Recognition
Jiafan Zhuang, Saihui Hou, Zilei Wang, Zheng-Jun Zha
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 |
|---|---|---|---|
| replace sliding-window recognition (Deep FCN) with semantic segmentation plus counting | license plate recognition · AOLP (average) | plate recognition accuracy (%): 97.90 -> 99.25 | Improved |
| replace character-detection-based pipeline (Hsu) with segmentation-based pipeline | license plate recognition · AOLP (average) | plate recognition accuracy (%): 86.93 -> 99.25 | Improved |
| use homogeneous random split instead of subset-based split | segmentation + counting LPR · AOLP | plate recognition accuracy (%): 99.25 -> 99.79 | Improved |
| simplify DeepLabv2 by keeping only the original-resolution branch (remove multi-scale branches) (ablation) | segmentation network · Segmentation backbone speed (GTX 1080Ti) | speed (FPS): 13 -> 38 | Improved |
| simplify DeepLabv2 by removing multi-scale branches (ablation) | segmentation network · LPR final accuracy | plate recognition accuracy | No clear effect |
| pre-train segmentation on 60,000 synthetic plate images before fine-tuning | segmentation model · license plate datasets | accuracy | Improved |
| add white lines to '1' and '7' images before counting (ablation) | counting refinement module · character counting | counting performance | Improved |
| add classifier on overlapped character areas (IoU > 0.5) for hard samples (ablation) | character sequence inference · license plate datasets | accuracy | Improved |
Closest papers by meaning
- Efficient License Plate Recognition via Holistic Position Attention (AAAI 2021 · similarity 0.88)
- Towards Ship License Plate Recognition in the Wild: A Large Benchmark and Strong Baseline (AAAI 2025 · similarity 0.86)
- LP-Diff: Towards Improved Restoration of Real-World Degraded License Plate (CVPR 2025 · similarity 0.84)
- SalM²: An Extremely Lightweight Saliency Mamba Model for Real-Time Cognitive Awareness of Driver Attention (AAAI 2025 · similarity 0.76)
- UPOCR: Towards Unified Pixel-Level OCR Interface (ICML 2024 · similarity 0.76)
- General Detection-based Text Line Recognition (NeurIPS 2024 · similarity 0.76)
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.
