🤖 PES Benchmark v0.2 for Detecting AI Motion
A tool has been introduced to identify AI-generated motion based on biometric entropy analysis. The method utilizes a 128-dimensional motion vector extracted from 3D pose sequences, achieving high separation accuracy (Cohen's d=10.4).
🌍 The emergence of specialized benchmarks to evaluate the "humanity" of motion is critical for combating deepfakes in video conferencing and secure identification systems. This establishes a new standard for Proof of Continuity.
👤 This technology enables the creation of more reliable ways to verify that a real human, rather than an advanced AI model, is on the other end of the screen.
Source 1: https://github.com/myshapeprotocol/myshape-protocol