Human Activity Recognition System
Human activity recognition (HAR) has become a popular topic of research because of its wide application. Computers are getting better at solving some very complex problems (like understanding an image) due to the advances in computer vision. Models are being made wherein, if an image is given to the model, it can predict what the image is about, or it can detect whether a particular object is present in the image or not. Here, a deep network architecture using residual bidirectional long short-term memory (LSTM) cells is proposed.Here we uses deep learning for Video Recognition - given a set of labelled videos, train a model so that it can give a label/prediction for a new video.Generally, the proposed network shows improvements on both the temporal (using bidirectional cells) and the spatial (residual connections stacked deeply) dimensions, aiming to enhance the recognition rate.Finally, the confusion matrix of the public domain UCI data set was analyzed.
Keywords - Human Activity Recognition; Bidirectional LSTM; Residual Network; K-Nearest Neighbor, Convolutional Neural Network