Ensure Safe Internet for Children and Teenagers Using Deep Learning

Abstract

Modern technology provides us with incredible resources that change how we live our lives daily. In today’s world, every single person uses a mobile phone. The child and teenagers also use mobile phones with the internet to communicate with their parents when they are in the office. Children and teenagers also use mobile phones for study, gaming, and social media. Sometimes the inappropriate content will appear before children and teenagers. Sometimes they cannot understand and click on it. We developed a proposed architecture based on CNN, RNN, OpenCV, haar cascade classifier, and MySQL for internet safety children and teenagers. When children and teenagers click on inappropriate content, the video camera will open and detect a child, teenager, adult, or old. If it is a child or teenager, the content will be hidden. OpenCV has been used for opening the video camera. Haar cascade classifier used for face detection. XAMPP MySQL database has been used for matching website links and blocking them. We generate a child, teenager, adult, and old(CTAO) dataset that contains 5000 images. The proposed architecture has been assessed using the CTAO dataset. We obtained 88.50% accuracy, 86.12% precision, 87.10% recall, and 86.60% f1 score.