Automatic Recognition of Human Actions

Understanding and recognizing human actions automatically have several applications in computer vision and human-machine interaction. In this paper, we aim to identify human actions in a short video automatically on KTH Dataset. We represented videos involving human actions into feature vector, whic...

Full description

Saved in:
Bibliographic Details
Published in2019 13th International Conference on Open Source Systems and Technologies (ICOSST) pp. 1 - 6
Main Authors Salahuddin, Sumayyea, Ahmed, Ibrahim, Rashid, Muhammad, Shafi-ur-Rehman, Minallah, Nasru
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.12.2019
Subjects
Cameras
Classifiers
Feature extraction
Feature Reduction
Human Action Recognition
Image edge detection
KTH Dataset
Legged locomotion
Self-organizing feature maps
Support vector machines
Training data
Online AccessGet full text

Cover

More Information
Summary:Understanding and recognizing human actions automatically have several applications in computer vision and human-machine interaction. In this paper, we aim to identify human actions in a short video automatically on KTH Dataset. We represented videos involving human actions into feature vector, which is invariant to small changes in the video and thus makes classification easier. We developed our model under reasonable assumptions that performs well. Our method achieves an average accuracy of 91.67% on 8 subjects with varying actions and gaits. We also compared our results with various algorithms and achieved good accuracy.
DOI:10.1109/ICOSST48232.2019.9043921