PTZJEPA: Predictive Active Vision for Autonomous Visual Data Collection

PTZJEPA is a closed-loop active perception framework that combines a predictive world model with reinforcement-learning control of a pan-tilt-zoom camera. The system learns predictive latent dynamics and uses them to select camera observations autonomously, allowing the camera to adapt its data acquisition strategy to the visual environment.

The project integrates image-based joint-embedding predictive learning (I-JEPA) with reinforcement-learning control inspired by Dreamer. It is designed for autonomous high-resolution scanning, environmental sensing, edge AI, and robotic perception tasks where field of view, bandwidth, or compute constraints make active sensing important.