Human Fall Detection and Tracking

Human fall is critical especially for elderly and small children, and human fall detection systems aim at detecting the falling activity as soon as possible. Fall detection can be done using wearable sensors. However, due to the requirement of frequent charging, discomfort, high cost, and potential side effects, wearable sensors are not always a suitable option for the elderly.

This project focuses on the basic research and investigation of vision-based fall detection and tracking algorithms based on Vision Transformers. Due to increased use of IoT solutions, there is an increasing trend of using cameras in airports, public places, and homes, and therefore, vision-based algorithms for fall detection are a better alternative from a futuristic point of view.

Team: Till Grutschus (TU Munich), Ola Karrar (Uppsala University) and Emir Esenov (Uppsala University)

Github Project        Paper


Early detection of human actions

Early detection of human actions is essential in a wide spectrum of applications ranging from video surveillance to health-care. While human action recognition has been extensively studied, little attention is paid to the problem of detecting ongoing human action early, i.e. detecting an action as soon as it begins, but before it finishes. This study aims at training a detector to be capable of recognizing a human action when only partial action sample is seen. To do so, a hybrid technique is proposed in this work which combines the benefits of computer vision as well as fuzzy set theory based on the fuzzy Bandler and Kohout’s sub-triangle product (BK subproduct). The novelty lies in the construction of a frame-by-frame membership function for each kind of possible movement. Detection is triggered when a pre-defined threshold is reached in a suitable way. Experimental results on a publicly available dataset demonstrate the benefits and effectiveness of the proposed method.

  • Ekta Vats and Chee Seng Chan, Early Detection of Human Actions – A Hybrid Approach, Applied Soft Computing, Volume 46, Pages 953-966, 2016. Paper
  • Ekta Vats, Chee Kau Lim and Chee Seng Chan, Early Human Actions Detection Using BK Sub-Triangle Product, In Proceedings of the IEEE International Conference on Fuzzy Systems (FUZZ-IEEE), Istanbul, Turkey, IEEE, Pages 1-8, 2015. Paper (Best student paper award nomination)
  • Chern Hong Lim*, Ekta Vats* and Chee Seng Chan, Fuzzy Human Motion Analysis: A Review, Pattern Recognition, Volume 48, Issue 5, Pages 1773-1796, 2015.     (* indicates equal contribution) Paper