Back to Search
Start Over
PhysGame: Uncovering Physical Commonsense Violations in Gameplay Videos
- Publication Year :
- 2024
-
Abstract
- Recent advancements in video-based large language models (Video LLMs) have witnessed the emergence of diverse capabilities to reason and interpret dynamic visual content. Among them, gameplay videos stand out as a distinctive data source, often containing glitches that defy physics commonsense. This characteristic renders them an effective benchmark for assessing the under-explored capability of physical commonsense understanding in video LLMs. In this paper, we propose PhysGame as a pioneering benchmark to evaluate physical commonsense violations in gameplay videos. PhysGame comprises 880 videos associated with glitches spanning four fundamental domains (i.e., mechanics, kinematics, optics, and material properties) and across 12 distinct physical commonsense. Through extensively evaluating various state-ofthe-art video LLMs, our findings reveal that the performance of current open-source video LLMs significantly lags behind that of proprietary counterparts. To bridge this gap, we curate an instruction tuning dataset PhysInstruct with 140,057 question-answering pairs to facilitate physical commonsense learning. In addition, we also propose a preference optimization dataset PhysDPO with 34,358 training pairs, where the dis-preferred responses are generated conditioned on misleading titles (i.e., meta information hacking), fewer frames (i.e., temporal hacking) and lower spatial resolutions (i.e., spatial hacking). Based on the suite of datasets, we propose PhysVLM as a physical knowledge-enhanced video LLM. Extensive experiments on both physical-oriented benchmark PhysGame and general video understanding benchmarks demonstrate the state-ofthe-art performance of PhysVLM.
- Subjects :
- Computer Science - Computer Vision and Pattern Recognition
Subjects
Details
- Database :
- arXiv
- Publication Type :
- Report
- Accession number :
- edsarx.2412.01800
- Document Type :
- Working Paper