Human activity recognition based on an amalgamation of CEV & SGM features

The method of marking video clips with action symbols is known as vision-based human activity recognition. Robust solutions to this problem have a variety of practical implementations. Due to differences in motion performance, recording environments, and inter-personal differences, the challenge is difficult. We specifically resolve these problems in this study work, and we solve imitations of state-of-the-art research. Projected human activity recognition is based on an amalgamation of CEV & SGM features. The proposed solution outperforms current models and produces state-of-the-art outcomes as compared to the best effectiveness of the control, according to experimental results on the datasets.