Year: 2025
Type: article
Abstract: This study focuses on the detection of real and deepfake faces in images and video data using the YOLO11 algorithm. Deepfakes, generated using advanced deep learning techniques, have the potential for misuse, including spreading false information and identity theft. The research employs public datasets, namely EC2-DeepFake and Deepfake Dataset, to train a YOLO11-based model. The performance of the model is evaluated using metrics such as mAP50 and mAP50-95. Results indicate moderate accuracy in distinguishing real from fake faces, with notable challenges in handling diverse data. The findings emphasize the need for improved training techniques and larger datasets to enhance detection performance. This work contributes to developing tools for mitigating the risks posed by deepfake technology.
Institution: Universitas Bina Sarana Informatika
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4407571246