Hand gesture recognition has emerged as an attractive and promising method in human-machine interaction. Applying the simple, efficient and inexpensive devices is necessary for this kind of the application. In this paper, we propose a novel hand gesture recognition by investigating the forearm muscles movement data processing sensed by an array of 8 Force Sensor Resistor (FSR). The acquired data is sent by the wireless device to the processing unit that is more convenient for users. The feature extraction scheme is proposed to get some useful information about the data. A graph of the FSR sensed signals are constructed by considering the extracted features from them. Weights of the graph edge are computed by finding the differences between correspondents pair of sensors. Multiobjective optimization is applied to find the optimum parameters and providing the best description of the sensors' relations in each class. The proposed approach will be evaluated by conducting experiments. 10 volunteered persons participated in gathering FMG data in different classes of the hand gestures. As the results show the proposed method has good performance to recognize hand gestures and has almost 93% accuracy in the overall.

Hand gesture recognition using force myography of the forearm activities and optimized features

Data of number of people in the room, calculations are usually carried out by assigning someone to oversee a room. In this final project, a system for calculating the number of people in the room is designed with image processing based on human detection that can be used in rooms, both for commercial applications and for security. This system uses Raspberry Pi device that already has an image processing method Haar-Cascade Classifier.   Input data is in the form of video taken directly via webcam to be captured into a frame so that it   can be used as a input the Haar-Cascade Classifier method and perform the counting process will be sent to the Antares platform. The system design has been tested with five scenarios. Scenario 1 the effect of the distance of the object, scenario 2 the effect of the pose of the object, scenario 3 the effect of the amount the object in the frame, scenario 4 affects the scale factor and scenario 5 measurement computation time. Scenarios 1 to 3 will do the best configuration for minimum neighbour. The system gets the best accuracy of 98,5% when the object distance 4 meters, the best accuracy of 96,6% when the object is facing forward and accuracy the best is 97,7% when the object in the frame is more than two objects with the best configuration use the minimum neighbour 5. Scenario 4 gets accuracy the best is 76,2% when using the scale factor 1.1. Scenario 5 gets the average computation time of the system is under one second, meaning the detection process done pretty fast.

/pdf/the-calculation-system-of-number-of-people-in-a-room-based-sqfot3miuy.pdf

The calculation system of number of people in a room based on human detection using haar-cascade classifier

/pdf/body-language-analysis-in-healthcare-an-overview-28sghx6j.pdf

Body Language Analysis in Healthcare: An Overview

The purpose of this research is to study and develop an alarm clock to help wake up at the set time. Without having to snooze the alarm and not be so sleepy that you have to go back to sleep again. to help reduce the risk of sleep-related disease By using various sensors, the Internet Of Thing (IoT) technology is used for health benefits. And from the evaluation of the accuracy of the system by experts. has a total value of 86% 

Smart Alarm Clock for Effective Sleep Health

Abstract A two stage real-time hand gesture recognition system is presented. It combines a machine learning trained detection step with a colour processing contour shape validation step. The detection step is done with either Adaboost Cascades or Support Vector Machines using HOG features. The system achieves a low false positive rate and a sufficient true positive rate necessary for robust real-time performance. It performs well compared to MobileNets a state of the art Neural Network for mobile real-time applications. 

/pdf/long-hands-gesture-recognition-system-2-step-gesture-38dnnnwt.pdf

Long Hands gesture recognition system: 2 step gesture recognition with machine learning and geometric shape analysis

This paper proposes to presents a hand gesture recognition as an assistive tool for patients care. The system uses Haar-like feature, Adaboost algorithm and Cascade Classification for hand detection. Principal Component Analysis work together with Histogram of Oriented Gradients and hand gesture decision to achieve preferable results of hand gesture recognition. Hand detection accuracy achieves 91.88 percent and sensitivity succeeds 95.75 percent.

Vassana Inmoonnoy

Papers

The message notification for patients care system using hand gestures recognition