Authors: Professor Swati Bagade, Dr. Anamika Jain, Aditya Sonune, Vedang Solaskar, Sanath Singh, Rishikesh Nate
Abstract: The widespread availability of altered images undermines confidence in digital media across multiple sectors. This work introduces an innovative web-based platform that harnesses deep learning to instantly distinguish genuine visuals from fabricated ones. Developed using Flask and a custom TensorFlow model with specialized components, the system allows users to upload images and receive rapid classifications through an intuitive interface. Experimental outcomes confirm its effectiveness in separating authentic from falsified content, supported by a design tailored for practical deployment. Unique activation functions and dropout techniques enhance the model’s durability, establishing it as a potent tool to counter visual misinformation.
