At AIST, visual computing covers the whole chain from image acquisition through analysis to visualisation. A prominent focus is drone-based wildlife monitoring: in BAMBI, Thumper and Pumbaa, wildlife and carcasses are detected, tracked and geo-referenced in aerial thermal and RGB footage. This work has produced, among other things, the publicly available BAMBI dataset and methods for generating synthetic training data.
In industry, the same methods are applied to automated quality inspection, for example AI-supported defect detection in non-destructive testing (FLARE) or turning technical drawings into 3D models and digital twins (WTA). This is complemented by work on mixed reality, motion analysis in sports (Ball-Tracking) and image generation (Stringnet).
