quang/ai
VI Hire me
Computer VisionBenchmark

Vietnamese License Plate Recognition Benchmark

A reproducible benchmark for Vietnamese automatic license plate recognition: three YOLO generations, three OCR pipelines, real RTSP traffic-camera data.

plate-level accuracy
98.9%
detector mAP@0.5
0.994
night-time frames accuracy
91%
detection images
8,259

Setup

  • YOLOv5, YOLOv8 and YOLOv11 trained under one 640×640, 20-epoch setup.
  • 8,259 detection images, a 3,763-sample recognition benchmark, three recognition architectures and 32 Vietnamese alphanumeric classes.

Detection

The best detector reached 0.994 mAP@0.5, 0.752 mAP@0.5:0.95, 0.995 precision and 0.985 recall on validation.

Recognition

A two-stage pipeline pairs YOLOv11 plate detection with a recognition model built from convolutional, recurrent and sequence-decoding (CTC) stages, reaching 98.9% plate-level accuracy.

Field conditions

Perspective-warp skew correction handles tilted plates, and the pipeline holds 91% accuracy on 15,000 night-time frames.

→ More work

Other case studies

BTS Digital Twin — Viettel AI Race 2026

Novel view synthesis of telecom (BTS) towers from drone captures with 3D Gaussian Splatting — then large urban scenes under an inference-time limit on an NVIDIA H200 in the final round.

 iPBMS — real-time patient monitoring

Fall and seizure detection on live RTSP hospital camera streams — from pose model to on-premise production on two cameras, with alerts to care staff in under two seconds.

 Inference-Efficient ConvMAE

Can a masked-autoencoder backbone be cheaper at inference without losing accuracy? Four hybrid designs, one fair protocol, measured under TensorRT rather than guessed from parameter counts.

 Face-Aware MAE

Identity-aware masked autoencoding: instead of masking random blocks, point MAE pre-training at the eyes, nose, mouth and central facial structure — and change nothing else.
esc