• Dileep Labana
• Dr. Kirit Modi

Recognizing human actions is a fundamental computer vision task with applications in video surveillance, HCI, and autonomous systems. This study examines the use of multi-task learning (MTL) methods to the detection of human action. MTL seeks to enhance the performance of individual activities by utilizing the shared knowledge among related tasks. The paper examines how MTL might improve the precision, robustness, and effectiveness of action recognition systems, reviewing the most recent advances in the field. The research looks at numerous MTL designs, such as collaborative training, shared feature learning, and task-specific regularization, and assesses how well they perform in various action recognition datasets. Additionally, the paper discusses the challenges and future directions in utilizing MTL for human action recognition, such as task selection, network architecture design, and dataset biases. By providing a comprehensive analysis of MTL-based approaches in human action recognition, this paper aims to contribute to the advancement of action recognition systems and inspire further research in this domain.

Multi Task Learning, Convolutional Neural Networks, Recurrent Neural Networks

"Human Action Recognition via Multi-Task Learning", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.10, Issue 7, page no.f409-f414, July-2023, Available :http://www.jetir.org/papers/JETIR2307548.pdf

"Human Action Recognition via Multi-Task Learning", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.10, Issue 7, page no. ppf409-f414, July-2023, Available at : http://www.jetir.org/papers/JETIR2307548.pdf