Modification of Deep Learning Technique for Face Expressions and Body Postures Recognitions

Abstract

Many modified techniques of deep learning systems for coma, are the most vital concerns that takes place amongst patients, mainly, for the elderly people and pregnant women, and cause mental and physical complications, which occur when individual falls-down automatically her/his head may hits the ground. The suitable distinguished medical treatments depend on the fast responses to the case; consequently, the fast detections of such issues are serious for such treatments. A modified method of deep learning system has been proposed in this paper, which can be considered as an effective and efficient tool for monitoring the patient, it also has many advantages. One of these advantages is that it does not depend on the person per se, but it depends on the surveillance cameras. The proposed system introduces a combination of some modern NN techniques, CNN and LSTM-RNN. Each technique in the proposed system was trained and evaluated to detect and classify human body postures (standing, sitting, lying, etc.) along with facial expressions (sad, happy, angry, etc.) with different standard datasets. And the full proposed structure has been offline and online trained and evaluated as one structure.