PhysicsAI awarded Phase II SBIR contract for Deep Reinforcement Learning for Model Training with Simulated Imagery

In this project PhysicsAI is developing a framework for using deep reinforcement learning (DeepRL) to create optimal training data from simulated imagery in order to train machine learning models with improved accuracy, increased robustness, and which support few/zero-shot detection for rare objects. In Phase II, we are developing the ability to use Generative Adversarial Networks (GANs), Auto-Encoders/Decoders, and DeepFakes to automatically create photorealistic synthetic backgrounds optimized by DeepRL agents to train object detectors for high-resolution satellite imagery.

SBIR Topic SCO182-006 is sponsored by the Office of the Secretary of Defense, Strategic Capabilities Office.