DeepTrailerAssist: Deep Learning Based Trailer Detection, Tracking and Articulation Angle Estimation on Automotive Rear-View Camera

Ashok Dahal, Jakir Hossen, Chennupati Sumanth, Ganesh Sistu, Kazumi Malhan, Muhammad Amasha, Senthil Yogamani; Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2019, pp. 0-0

Trailers are commonly used for transport of goods and recreational materials. Even for experienced drivers, manoeuvres with trailers, especially reversing can be complex and stressful. Thus driver assistance systems are very useful in these scenarios. They are typically achieved by a single rear-view fisheye camera perception algorithms. There is no public dataset for this problem and hence there is very little academic literature on this topic. This motivated us to present all the trailer assist use cases in detail and propose a deep learning based solution for trailer perception problems. Using our proprietary dataset comprising of 11 different trailer types, we achieve a reasonable detection accuracy using a lightweight real-time network running at 30 fps on a low power embedded system. The dataset will be released as a companion to our recently published dataset [24] to encourage further research in this area.