Facebook is 65% capable of detecting Deepfake videos

news image

The Facebook authorities did not get the expected results by organizing political propaganda, pornography, or aggressive fake videos 'Deepfake Detection Challenge' using deep learning technology. This is because the high-performance models have been able to detect a maximum of 75.18 percent of Deepfake video accurately.

Two thousand people participated in the competition. During this time, they received training and created a new dataset by creating about 35,000 different models. They work with one lakh videos of 3,000 actors and actresses. Among them, Selim Seferbekov identified fake videos with an average of 85.18 percent, ranking first, WM second, Enteclab third, Eighteen Years Old fourth and The Medics fifth.

In a statement, Facebook authorities said the competition created a baseline for identifying deepfake videos. This competition will play an important role in creating a large number of different data sets and benchmarking dipfeck detection models, working with new technologies and learning from each other.