Weigh stations, recycling centers, and waste management facilities rely heavily on accurate vehicle identification and weight tracking to process transactions, enforce regulations, and maintain operational efficiency. They process hundreds of trucks daily, 24 / 7, continuously.
Current systems rely on human-verification, as they are fully manual. Manual steps create bottlenecks and limit efficiency, which can then lead to errors in data entry, affecting compliance and record accuracy.
What this project aims to do is automate the entire vehicle weight process: from vehicle identification to cargo classification, all the way to vehicle weight tracking and data storage. Our Vehicle Weight System creates a solution that provides a safer work environment, as well as a much more efficient workflow regarding truck processing.
TensorFlow, YOLO, OCR, LPR model and Cargo Detection Model
This project uses a License Plate Recognition (LPR) model to conduct vehicle identification, tracking and storing the license plate. This model uses YOLO for plate image detection and EasyOCR for text extraction from plate image. This project also uses an image classification model MobileNet, in combination with YOLO, to detect what type of cargo each truck has
Python, Flask, PostgreSQL, Docker
The backend for this project primarily uses Python. It uses Flask as a proxy server for our multi-threaded tracking approach, also using PostgreSQL for our database Docker to wrap everything inside
Jetson Orin Nano, Cameras, OpenCV
This entire project is hosted on a Jetson Orin Nano by NVIDIA. This type of board is perfect for this project, as the board is specifically made with AI and machine learning models in mind. We also use two Innomaker cameras: one for the LPR detection model and another for the Cargo Detection model.
Project Manager / LPR Developer
Backend / Database Developer
Backend Developer
Cargo Detection Developer
Hardware / Sensor Engineer
