License Plate Recognition on PKU
99.5Recognition RateTRBA + CCPD’s training set + synthetic
Evaluation results
| Method | Links | |
|---|---|---|
| TRBA + CCPD’s training set + syntheticReal images of Chinese LPs used for training=100,000+, Multinational=true2026.01 | 99.5 | |
| Li et al.Latency (ms)=50, Hardware=FPGA2026.06 | 99.5 | |
| TRBA + syntheticReal images of Chinese LPs used for training=506, Multinational=true2026.01 | 99.4 | |
| Wang et al. (2022b)Real images of Chinese LPs used for training=100,000+, Multinational=false2026.01 | 96.9 | |
| TRBAReal images of Chinese LPs used for training=506, Multinational=true2026.01 | 96.8 | |
| TRBA + CCPD’s training setReal images of Chinese LPs used for training=100,000+, Multinational=true2026.01 | 96.8 | |
| Zou et al. (2020)Real images of Chinese LPs used for training=100,000+, Multinational=false2026.01 | 96.6 | |
| Rao et al.Real images of Chinese LPs used for training=4,444, Multinational=false, Applied authors code and pre-trained models=true2026.01 | 96.1 | |
| OpenALPRReal images of Chinese LPs used for training=?, Multinational=false2026.01 | 96 | |
| LPCRLatency (ms)=4, Hardware=FPGA2026.06 | 94.2 | |
| Chen et al.Real images of Chinese LPs used for training=100,000+, Multinational=false2026.01 | 92.8 | |
| Fan & ZhaoReal images of Chinese LPs used for training=100,000+, Multinational=true2026.01 | 92.5 | |
| Zhang et al.Real images of Chinese LPs used for training=100,000+, Multinational=false2026.01 | 90.5 | |
| SighthoundReal images of Chinese LPs used for training=?, Multinational=true2026.01 | 89.3 | |
| Zhang et al.Latency (ms)=25, Hardware=GPU2026.06 | 88.2 |