Deep Metric Learning Framework for Vehicle Detection and Tracking
DOI:
https://doi.org/10.17762/msea.v71i4.905Abstract
The identification of vehicles is a challenging task; the gathered information may produce the deviation from mathematical computations. The deep metric learning framework is proposed to detect and classify the vehicles; the proposed model has the process of bounding box construction, vehicle tracking and detection. The embedding vector is used to process the input images and fed into a convolutional neural network for obtaining the vehicle tracking process with the computed distance within the vehicles and the measurement of the similarity degree between the vehicles. BIT Vehicle Dataset is involved for testing the proposed framework; the dataset has the different kind of vehicle images for performing the evaluation process.