Undergraduate research at the Autonomous Robot Intelligence Lab (ARIL). I reconstruct objects from 2D images with Instant-NGP, render novel 2D views from the 3D model, auto-annotate the rendered views, and train a YOLO detector on the augmented data (for example, quadruped-robot detection).
Pipeline: Instant-NGP to novel-view rendering to auto-annotation to YOLOYOLO detection trained on the augmented data
