• Deepak Patel
• Aman Patel
• Aditya Choudhary
• Nidhi Nigam
• Shruti Lashkari

Deepfake technology, driven by powerful deep learning techniques, has be- come a serious concern due to its potential misuse in misinformation, fraud, and cyber- crime. This study proposes a hybrid model combining ResNeXt CNN for extracting spatial features from video frames and LSTM for analyzing temporal sequences. The system was trained using well-known datasets—FaceForensics++, Celeb-DF, and the Deepfake Detec- tion Challenge—and achieved high accuracy in identifying manipulated media. Our ap- proach not only outperformed traditional models like XceptionNet but also demonstrated robustness across different types of deepfake techniques, making it suitable for practical deployment in media verification and forensic investigations.

Deepfake detection; ResNeXt CNN; LSTM; Video Forensics; AI-generated media; Temporal analysis

"Unmasking Deepfakes: A Hybrid ResNext CNN-LSTM Approach for Robust Video Authentication", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.12, Issue 5, page no.13-17, May-2025, Available :http://www.jetir.org/papers/JETIRGV06004.pdf

"Unmasking Deepfakes: A Hybrid ResNext CNN-LSTM Approach for Robust Video Authentication", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.12, Issue 5, page no. pp13-17, May-2025, Available at : http://www.jetir.org/papers/JETIRGV06004.pdf