Interdependent Multi-task Learning for Simultaneous Segmentation and Detection

Mahesh Reginthala, Yuji Iwahori, M. Bhuyan, Yoshitsugu Hayashi, Witsarut Achariyaviriya, Boonserm Kijsirikul

2020

Abstract

Lightweight, fast, and accurate deep-learning algorithms are essential for practical deployment in real-world use-cases. Semantic segmentation and object detection are the principal tasks of visual perception. A multi-task network significantly reduces the number of parameters compared to two independent networks running simultaneously for each task. Generally, multi-task networks have shared encoders and multiple independent task-specific decoders. Instead, we modeled our network to exploit the features from both encoder and decoder. We propose the multi-task network that performs both segmentation and detection with only 37.9 million parameters and inference time of 74 milliseconds on a consumer-grade GPU. This network performs two tasks with much fewer parameters and in much less inference time compared to each single task network.

Paper Citation

in Harvard Style

Reginthala M., Iwahori Y., Bhuyan M., Hayashi Y., Achariyaviriya W. and Kijsirikul B. (2020). Interdependent Multi-task Learning for Simultaneous Segmentation and Detection. In Proceedings of the 9th International Conference on Pattern Recognition Applications and Methods - Volume 1: ICPRAM, ISBN 978-989-758-397-1, pages 167-174. DOI: 10.5220/0008949501670174

in Bibtex Style

@conference{icpram20,
author={Mahesh Reginthala and Yuji Iwahori and M. Bhuyan and Yoshitsugu Hayashi and Witsarut Achariyaviriya and Boonserm Kijsirikul},
title={Interdependent Multi-task Learning for Simultaneous Segmentation and Detection},
booktitle={Proceedings of the 9th International Conference on Pattern Recognition Applications and Methods - Volume 1: ICPRAM,},
year={2020},
pages={167-174},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0008949501670174},
isbn={978-989-758-397-1},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 9th International Conference on Pattern Recognition Applications and Methods - Volume 1: ICPRAM,
TI - Interdependent Multi-task Learning for Simultaneous Segmentation and Detection
SN - 978-989-758-397-1
AU - Reginthala M.
AU - Iwahori Y.
AU - Bhuyan M.
AU - Hayashi Y.
AU - Achariyaviriya W.
AU - Kijsirikul B.
PY - 2020
SP - 167
EP - 174
DO - 10.5220/0008949501670174