Action Analyzer for Differently Abled People

This paper presents a simulation system for differently abled people that utilizes the Long Short-Term Memory (LSTM) algorithm and the Mediapipe Holistic Model (MHM). The system aims to enhance the daily lives of differently abled individuals by analyzing their body movements and translating them into meaningful actions. The Mediapipe Holistic model extracts key points of the movements from input video frames, and the LSTM algorithm processes the resulting data to recognize various actions such as scared, understand, Wednesday. The system is trained on a diverse set of images to ensure accuracy and robustness in recognizing a wide range of actions. Results show that the combination of MHM and LSTM provides a high accuracy model even with less dataset in comparison to many models proposed over the years for action analyzer system. The model has potential for use in assistive technologies for differently abled individuals. The model shows 92^% accuracy on the training dataset and 90% accuracy on the testing dataset. Results show that the proposed system outperforms existing approaches and has potential for use in assistive technologies for differently abled individuals.