EE Seminar: RadarBOT-SORT: Integrating Kalman Filter and Neural Network for Multi-Object Tracking by Radar
https://us05web.zoom.us/j/89308496750?pwd=CZmbhbnb9cZRRpIfG1MPeZ5INBdzX1.1
Electrical Engineering Systems ZOOM Seminar
Speaker: Idan Daniel
M.Sc. student under the supervision of Prof. Ben-Zion Bobrovsky
Wednesday, 9th July 2025, at 15:00
RadarBOT-SORT: Integrating Kalman Filter and Neural Network for Multi-Object Tracking by Radar
Abstract
Multi-object tracking in radar systems faces unique challenges due to measurement noise, detection confidence uncertainty, and the absence of rich visual features available in camerabased systems. This seminar presents RadarBOT-SORT, a novel tracking framework that integrates neural network confidence scores with Kalman filter-based SORT (Simple Online Realtime Tracking) algorithms for enhanced radar object tracking performance.
The research addresses a fundamental question in radar signal processing: whether postprocessing deep learning detection outputs with tracking algorithms improves overall system performance compared to raw neural network predictions. The proposed system combines a deep learning network to discover detections in radar Range-Azimuth-Doppler outputs for radar object detection, with a confidence-enhanced SORT (Simple Online Realtime Tracking) algorithm that incorporates detection confidence scores into both track association and Kalman filter measurement noise adaptation.
Idan Daniel is an M.Sc. student and deep learning engineer with experience in radar deep learning algorithms development, specializing in AI applications for radar systems.
השתתפות בסמינר תיתן קרדיט שמיעה = עפ"י רישום בצ'ט של שם מלא + מספר ת.ז.
