Explainability in AI-based CV from development to deployment


AI-based approaches for signal processing often have the character of a black box. AI explainability is all about providing tools which help answer two fundamental questions arising from that: Why did the system create the observed output? How can the user influence the AI-based system? We will discuss key aspects of explainability of AI-based systems for object-detection in safety-critical environments from a user‘s as well as a developer’s perspective.