The UTC Graduate School is pleased to announce that Yasir Hassan will present Master's research titled, Deep Learning-Based Framework for Traffic Estimation on 07/06/2023 at 11 am in Multi-Disciplinary Research Building - Auditorium. Everyone is invited to attend. 

Computer Science

Chair: Mina Sartipi

Abstract:
In this work we introduced a deep learning-based framework for vehicles detection, tracking and speed estimation. YOLOv7 has been trained to detect and classify vehicles into Sedan, SUV, Pickup truck and Bus with a reported MAP of 0.69. For vehicles re-identification, we investigated the use of SORT with A Siamese network as its deep feature extractor (revision of DeepSort). Both the Siamese network and the CNN of the DeepSort was trained on UA-DETRAC dataset. The two models have been tested on the KITTI dataset and the revised model showed an average reduction rate of 71% in the IDSW rate when compared to DeepSort. A total of 11 cameras have been calibrated assuming a pinhole model to estimate camera parameters, then through perspective transformation, reference object scaling and zones creation, we were able to estimate vehicles’ speeds with an error rate of 0.516 mph. The amount of rich and accurate data our framework produce paves a way for different potential application, such as travel time estimation for multi-camera tracking and creating speed data benchmarks.