Research papers · Posted 2024.04.08 · Preprint
CLIPping the Limits: Finding the Sweet Spot for Relevant Images in Automated Driving Systems Perception Testing
CLIPping the Limits: Finding the Sweet Spot for Relevant Images in Automated Driving Systems Perception Testing
Overview
The method automatically selects relevance thresholds for image retrieval driven by natural-language prompts. It balances false positives and missed relevant images, with a fallback when assumptions fail, supporting automated-driving perception test-data selection.
Version: v2