• B. Ankitha
• D. Mercy Sunada
• Y. Bhargavi
• Y. Manasa
• T. Bhaskar

We provide a novel method for detecting synthetic face modifications in photographs using deep learning techniques in light of the growing threat posed by deepfake images. Our approach is designed to tackle problems caused by AI-generated deepfakes; specifically, it focuses on distinguishing real from fake facial images. A ResNetV1.5 Convolutional Neural Network (CNN), which is skilled at extracting complex features from facial data, is at the heart of our system. These characteristics are then used to train a recurrent neural network (RNN) of the Long Short-Term Memory (LSTM) type for image-level classification, which allows it to discern between real and fraudulent information. In order to verify the stability and practicality of our model, we carry out thorough tests with the VGGFace2 dataset, which is a well-known tool for face recognition studies. Our approach involves combining multiple datasets, such as FaceForensic++, Deepfake Detection Challenge, and Celeb-DF, to create a balanced and diversified dataset that is representative of real-world situations. Moreover, our approach prioritizes efficiency and simplicity, enabling competitive performance in the detection of altered imagery. Our study uses AI technology to tackle the urgent problem of the spread of deepfakes. By combining an LSTM-based RNN with a ResNetV1.5 CNN, we are able to identify fake face modifications with impressive accuracy. Our thorough analysis of a variety of datasets highlights the effectiveness and usefulness of our strategy in halting the spread of deepfake images in practical contexts.

Synthetic face modification, CNN, LSTM, Deep learning, MTCNN

"Deepfake Detection", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.11, Issue 4, page no.e498-e502, April-2024, Available :http://www.jetir.org/papers/JETIR2404454.pdf

"Deepfake Detection", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.11, Issue 4, page no. ppe498-e502, April-2024, Available at : http://www.jetir.org/papers/JETIR2404454.pdf