Analysis of Employee Surveillance System using Deep Learning Models

Achieving maximum productivity becomes difficult in the conditions when the corporate working arrangements change from office to home. Any type ofinattentiveness can cause companies to incur huge financial losses. So in order to increase work efficiency and lessen the errors and risks in work, there comes a need for a system that monitors an employee's performance. This research paper titled “Analysis of Deep Learning Models for Employee Surveillance System” proposes an idea that will help corporate employers to solve this issue by providing a system that has a feature of monitoring attentiveness of their remote employees. This system is developed using OpenCV for real-time video capturing and deep-learning algorithm CNN for image classification and haar-cascade for cascading. CNN algorithm has been used in order to classify 3000 images of various eye states in a more proficient manner than traditional ones. It alerts the employee by an alarm which starts triggering after a certain threshold and it also starts recording a video by the same camera which will act as an evidence of the employee's negligence to his work. In addition, an internet checker has been installed, which allows it to analyze the internet connectivity of employee machines. Additionally, a SMS API has been used to notify the company/employer of the status of inactivity of the user/employee using a message or notification. A variety of models are being tested to determine the accuracy of the system. The conclusion of analysis work has been derived in order to come up with a more efficient system.