Sign Language Detection using LSTM Deep Learning Model and Media Pipe Holistic Approach

In this study, we aim to develop a deep learning model to recognize alphabet signs depicted in images and videos. We collect hand movement data using media pipe, which extracts coordinates of 21 points on the palm. These coordinates are then converted into a NumPy array and input into a long short-term memory (LSTM) model for sign detection. The output of the model is a string representation of the detected alphabet, ranging from 'A' to 'Z'. To improve the scientific validity of the model, we plan to use a larger and more diverse dataset, conduct cross-validation, and perform statistical analysis of the results. We will also consider using alternative deep learning models and data preprocessing techniques to further optimize model performance.