An Efficient Algorithm for Fake Video Detection
Nafi Nibras, Shazzadul Fahim, Sadman Sakib, Shiam Ur Rashid, Atiqur Rahman · 2024
In recent times, the circulation of fake videos has caused significant uproar among the general people. Fake videos have caused considerable inconvenience in the daily lifestyle of some people. The intervention of fake videos is now regarded as a serious obstacle due to its manipulating and convincing power. Researchers and scientists have been trying to implement different methods and techniques to overcome this new technical barrier. Various procedures and algorithms have been implemented, which showed a wide variety of results in controlling and defeating this technological hazard. However, it is yet to be concluded on which method should be followed to stop fake videos from causing us problems. This is due to the broad range of results achieved by them and the deficiencies that these methods contain. None of these methods produced the perfect protection in all aspects of fake videos. Hence, our project emphasizes on detecting fake videos by implementing algorithms that will provide the most efficient and accurate result. In our project, a video will be uploaded in the web application, which will then undergo different machine learning techniques which will involve filtering, feature extraction, and classification. Finally, after completing the procedure, the output will be displayed whether the video is fake or real. According to our model, the accuracy rate achieved was 83.456%, 87.349% and 88.965% for the length of sequence 10, 20 and 40 respectively. The best accuracy was achieved when the length of sequence is 40.