TempGlitch: Evaluating Vision-Language Models for Temporal Glitch Detection in Gameplay Videos
Researchers have introduced TempGlitch, a new benchmark designed to evaluate how well vision-language models (VLMs) can detect temporal glitches in gameplay videos. Unlike previous methods that focused on static visual anomalies, TempGlitch specifically tests the models' ability to identify issues that only become apparent when observing changes across sequential frames. Initial evaluations of 12 different VLMs revealed that current models perform poorly, often struggling to distinguish between actual glitches and normal gameplay, indicating a significant gap in their temporal reasoning capabilities. AI
IMPACT Highlights a critical gap in current vision-language models' ability to understand temporal dynamics, potentially guiding future research in AI for game quality assurance.