As one of most active research areas, computer vision and deep learning has been used for human identification and human activity recognition. Under the H-MASINT program (a multi-year, multi-million dollars cross directorate program charted by the AFRL), Innovision’s chief scientist had been leading a group of scientists and engineers utilizing computer vision and deep learning to develop technologies for human identification and human activity recognition. In recent years, using deep learning, Innovision has built technologies that can extract 3-D human shape model from 2-D imagery, estimate body measurements from a body shape model, and crate a shape model based on body measurements.

Click on the video to the right to for the demo video of the PASSIG system developed by Innovision.