IDENTIFYING SUSPICIOUS ACTIVITIES ON TWITTER USING NLP AND COMPUTER VISION

Human trafficking, a global crisis, often spreads online through secret messages on social media. Our approach uses NLP to detect suspicious tweets promoting illicit services and child abuse. Tweets with child-related hashtags are mined in real time. Haar models identify facial and body geometry, focusing on torso-to-head proportions even with blurred faces. Images and URLs are classified, and an SVM model using torso features outperforms CNN for detection.