Skip to content

Improved Computer Vision-based Framework for Electronic Toll Collection

Notifications You must be signed in to change notification settings

derinben/Vehicle-ReIDentification

Repository files navigation

Improved Computer Vision-based Framework for Vehicle Re-id

This model proposes an automatic vehicle fingerprinting system that helps quick re-identification of adversarial vehicles.

The VeRi776 dataset which contains real-world vehicle images taken from surveillance cameras and is used to train the models involved behind the research of vehicle re-identification.

With this, a combination of license plate recognition and vehicle re-identification techniques is used in the proposed framework to improve the efficiency of identifying vehicles using the cameras in toll plazas

The proposed framework employs Siamese model architecture to identify the attributes such as color, model, and type of vehicle to re-identify the same vehicle. We have subclasses the Data Generator API to create our own data generator for training the Siamese Model along with a script to generate the positive and negative pairs.

Releases

No releases published

Packages

No packages published