Automated Child Daycare Monitoring System-A Deep Learning Method To Detect Unsafe Behaviour Upon Children

Abstract

Abuse in daycare centers is a very serious problem in modern parenting. Our lives are becoming busier every day as we place a greater emphasis on our professional lives. Since most parents choose to send their kids to daycare while they are at work, there have been a lot of recent reports of child abuse at daycare. Every year, there are more abused kids in daycare than the year before, and sometimes this results in significant injuries. Daycare is now seen by parents as a risk to their kids. As a result, the daycare industry is ultimately failing, and many parents' jobs will be hampered by the issue of child care. Therefore, a method that can instantly classify child abuse in daycare using CCTV footage must be developed. I decided to create a mechanism that will aid in the authority's classification of child abuse detection as a result. To find child maltreatment in daycare, I employed image processing and a custom convolutional neural network method. The method produced a promising result with an astonishing 84% accuracy in this case. In order to improve accuracy and incorporate this in CCTV to receive notifications of abuse in real-time, I will work on a large dataset with more Transfer learning techniques in the near future.