The KITTI Vision Benchmark Suite

DEFT is joint model of detection and tracking. Our approach relies on
an appearance-based object matching network jointly-learned with
an underlying object detection network. An LSTM is also added to
capture motion constraints.

TBD

@InProceedings{Chaabane2021deft_2021_CVPR_Workshops,
author = {Chaabane, Mohamed and Zhang, Peter and Beveridge,
Ross and O'Hara, Stephen},
title = {DEFT: Detection Embeddings for Tracking},
booktitle = {Proceedings of the IEEE/CVF Conference on Computer
Vision and Pattern Recognition (CVPR) Workshops},
month = {June},
year = {2021}
}