Overview
ScaryNet Filter is a demonstration platform showcasing the capabilities of ScaryNet, a deep learning-based convolutional neural network (CNN) designed to detect and filter images containing potentially disturbing or scary content. This web application utilizes only HTML5 and JavaScript to ensure portability across all modern platforms and devices.
The application works entirely in the browser and can run on any device with an HTML5-compliant web browser, including desktops, laptops, mobile devices, smart TVs, and Internet kiosks.
What is ScaryNet?
ScaryNet is a compact artificial convolutional neural network (CNN) trained to identify and filter images that may be considered scary or disturbing. The model is optimized to be fast and lightweight, enabling real-time performance even on resource-constrained environments such as mobile browsers or embedded devices.
ScaryNet is trained with TensorFlow and Keras, and is deployed using the TensorFlow.js library for seamless execution within the browser environment.
Use Cases
- Automatic image filtering for websites or web applications.
- Protecting children from exposure to scary images and media previews.
- Identifying and labeling scary images or content in multimedia collections.
- As a plugin or module for platforms such as WordPress, Joomla, Google Chrome, and others.
Author
Mohammad Hafiz bin Ismailmypapit@gmail.com
For academic or professional enquiries: deeplearn@uitm.edu.my
Citing this Work
Download a BibTeX reference file or click DOI for citation:
