Abstract:
In day-to-day existence, people act on many tasks. It is vital to record and analyze the daily presence of individual people. Hence it could assist with relieving a few medical conditions and different issues. Human Activity Recognizing is a key component research topics in computer vision for different sectors like security monitoring, healthcare and human-computer association, and sports. Nowadays, the smartphone has become popular and helpful for people. Because smartphone has many various and effective sensors, in this paper, we have used smartphone sensors: an accelerometer and gyroscope to detect human activity. In our research, we collected 30 study participants labeled datasets between ages nine-teen to four-ty-eight (19-48) years who have executed actions such as activities of daily life include sitting, walking, standing, walking up or down stairs, and lying down while using a smartphone equipped with such sensors. The objective is to do each of the six activities in the correct order. Two sets of the record dataset were randomly chosen, with 70% of participants 30% were chosen to produce test data, the remaining 70% to produce training data. The results were gained along with compared by supervised learning algorithms like Decision Tree Classifier, Random Forest, K Nearest neighbor method, Logistic Regression, and Support Vector Machines algorithms. By comparing those algorithms, we gained the best results accuracy from Logistic Regression which is 96.21%.
