Developed by ARSA Technology, this project is a lightweight, flexible, and efficient license plate recognition system tailored for deployment on a range of devices. It offers multiple inference backends including CPU, CUDA (GPU), and OpenVINO for edge computing.
- ✅ Real-time License Plate Detection
- 🧠 Multiple Inference Backends:
arsalpr_cpu.py: for CPU-only environmentsarsalpr_cuda.py: for NVIDIA GPU acceleration (CUDA)arsalpr_vino.py: for Intel OpenVINO toolkit
- 🔁 Client-server architecture for modular integration
- 🧪 Sandbox mode for custom algorithm testing
Simple-AI-ALPR-main/
├── arsaLpr/
│ ├── arsalpr_cpu.py
│ ├── arsalpr_cuda.py
│ ├── arsalpr_vino.py
│ ├── client.py
│ ├── server.py
│ ├── sandbox_algorithm.py
│ └── assets/ (model weights and config)
├── setupvars.sh
git clone https://github.com/arsa-technology/Simple-AI-ALPR.git
cd Simple-AI-ALPRMake sure Python ≥3.8 is installed.
pip install -r requirements.txtRun with CPU:
python arsaLpr/arsalpr_cpu.pyRun with CUDA:
python arsaLpr/arsalpr_cuda.pyRun with OpenVINO:
python arsaLpr/arsalpr_vino.pyModel configs and weights are stored under:
arsaLpr/assets/
These include:
- YOLOv3-tiny based configuration for plate detection
- Custom-trained weights for Indonesian license plates
- ESPCN super-resolution model
To experiment with custom algorithms, modify:
arsaLpr/sandbox_algorithm.pyThis project is © 2025 ARSA Technology. All rights reserved. For licensing inquiries, please contact us at arsa.technology/contact.
ARSA Technology is an Indonesia-based deep tech company developing cutting-edge AI and IoT solutions for smart infrastructure, security, and automation.
Visit us at: https://arsa.technology