There are usually two types of methods to deal with and detect deep face forgery. One group uses deep learning to automatically extract important and distinctive features such as eye blink rate, head and neck movement, eyebrow movement, facial skin furrows, lip and mouth movement, and shadows in order to detect deep forgery. Another category of deep face forgery detection methods are based on video cryptography techniques (such as video encryption, authentication codes, tagging, abstraction, and digital signature). Sometimes the encryption algorithm is applied to secure the video.
Each of the existing methods for detecting deep facial forgery has weaknesses and shortcomings. Unfortunately, the high error rate in these methods raises these concerns about the fundamental need for more powerful approaches to distinguish fake from real videos in the future. Also, it is necessary to use expensive computational tools to detect deep forgery due to the large amount of calculations. Methods based on video encryption also face several challenges.