Detection of Real vs. Fake Face Enhanced by MobileNetV2
Vijay Madaan, Neha Sharma · 2025
This study's abstract inspects the MobileNetV2 architecture and discovers that it performs excellently in accuracy and loss metrics over 20 training sessions. After the 20th session, the accuracy has been increased from 54.83% to 90.59% whereas the loss has been decreased from 1.1261 to 0.2207. This advancement improves security protocols, reduces identity theft, dismisses false information, supports debatable investigation, privacy protection, and confidence promotion in digital material. These consequences show the potential to discriminate between real and altered face photographs that are enhanced through increased consistency between its predictions and actual data.