A Human Activity Determination Predicting Abnormality Using SVM Approach for Mining Field Workers

This paper presents a human activity determination system for predicting abnormalities in mining field workers employing a Support Vector Machine (SVM) approach. This system integrates various sensors including an oxidation sheet sensor, temperature sensor and pulse oximeter sensor to capture the data related to worker activities. The collected data is processed by an ESP32 controller programmed with a classifier, which is capable of learning patterns and predicting the occurrence of abnormal activities based on the sensor readings. The predicted results are then transmitted to a LoRa transceiver for wireless communication. The LoRa receiver receives the transmitted information and relays it to a NodeMCU, which serves as a bridge between the receiver and the mobile application. The mobile application is designed to display the oxidation data, heart rate data, and temperature data obtained from the sensors in real-time. By analysing the sensor data and utilizing the SVM classifier, the system aims to detect and alert mining field workers to potential abnormal activities, thus enhancing safety measures and minimizing the risk of accidents in mining environments. The proposed SVM classifier has a specificity of 89.47%, sensitivity of 89.47 and accuracy of 92.64%.