Deepfake Video Detection using CNN based Architectures and Vision Transformer Model

Author(s)	Ms. Sumedha Arya
Country	India
Abstract	Deepfake videos are becoming more advanced and realistic, that their detection is getting more challenging. It brings a deep concern for privacy and security on digital platforms. Current techniques perform well in closed environment but faces issues in real world scenario for deepfake detection. Therefore, in this work, we focus on detecting deepfake videos by first breaking each video into individual image frames and then analyzing those frames using deep learning models. We applied transfer learning using various CNN based pretrained models such as VGG, ResNet, DenseNet, MobileNet, EfficientNet along with Vision Transformer (ViT). Our results show that all models perform well, with accuracy values between 92.56% and 97.16% for CNN-based models. The Vision Transformer performs the best overall, achieving 99.00% accuracy by capturing global patterns and small manipulation details in deepfake videos. Overall, the study proves that transfer learning, is a strong and reliable approach for detecting deepfake videos even when the size of the dataset is less.
Keywords	Deepfake Detection, CNN, Vision Transformer, Transfer Learning, Video Frame Analysis
Published In	Volume 17, Issue 1, January-March 2026
Published On	2026-02-11

About IJSAT Fees & Payment Current Issue Publication Archive	Submit Research Paper Track Submission Status Publication Guidelines Publication Ethics Peer Review & Plagiarism	Join as a Reviewer Editors & Reviewers Reviewer Referral Program Get Reviewer Membership Certi.	Website/Journal Policies Usage Policy Content Policies Privacy Policy

Contact Us	Message on WhatsApp	+91-9687-182-185	editor@ijsat.org

International Journal on Science and Technology