Suspect Detection in Real Time using Generative AI Generated Dataset

Real-time suspect detection presents a formidable challenge for AI and ML methods due to the need for extensive image data. This work proposes a solution using Generative AI algorithms like GAN, StyleGAN, and TGAN to create a diverse high-definition test dataset from a single suspect image. Through feature-guided face manipulation, the algorithms generate realistic facial expressions and scenarios, expanding the dataset without requiring numerous unique images. Subsequently, the augmented dataset trains a Face Detection model, which analyses live video streams to identify suspects. Upon detection, the system triggers realtime alerts for swift intervention. By leveraging Generative AI for dataset augmentation and training robust Face Detection models, the project aims to improve suspect detection accuracy and efficiency in real-world settings, bolstering public safety measures.