Advanced Cyberbullying Detection: Integrating Pytesseract, Demoji, and BERT for Comprehensive Textual and Visual Content Analysis

In recent years, Cyberbullying on social media platforms leads to serious problems among children such as mental and health issues. To overcome these issues an advanced approach for cyberbullying detection and analysis method is needed. Traditional methods of cyberbullying detection focus mainly on text, often overlooking the significance of visual elements such as images and emojis. While recent techniques leveraging Natural Language Processing (NLP) and Computer Vision have shown promise, they face several challenges. Demoji is utilized to decode and normalize emojis, which are frequently used to convey harmful messages. Additionally, emoji decoding tools like Demoji provide textual representations of emojis but it lack in the ability to infer their contextual intent. These limitations are further compounded by challenges such as detecting sarcasm, handling cultural variations, and processing real-time data at scale. To enhance the cyberbullying detection, an effective method is used by integrating three powerful tools such as Pytesseract, Demoji, and BERT, applied to textual and visual content. Experimental results demonstrate the effectiveness of this integrated methodology, showing improved detection accuracy compared to traditional text-based approaches. This system offers a promising step toward more robust and multi-faceted cyberbullying prevention strategies.