SemComp-Bench：视频生成中语义任务完成的基准测试

AI 摘要

该论文提出了语义任务完成视频生成任务，要求生成视频既实现预期结果又具备语义基础。作者构建了包含六个领域的SemComp-Data数据集，并设计了基于视觉语言模型的SemComp-Bench评估协议，通过OA分数和GR分数分别衡量结果达成和生成可靠性。实验表明，现有视频生成模型在实现预期结果和保持语义基础方面仍面临挑战。

正文节选

SemComp-Bench: Benchmarking Semantic Task Completion in Video Generation Abstract Semantic task completion video generation evaluates whether generated videos achieve intended outcomes with semantic grounding, supported by a curated dataset and vision-language model-based benchmark. We introduce Semantic Task Completion Video Generation, an outcome-oriented video generation task. Under this formulation, success requires both achievement of the intended outcome and semantic grounding. Semantic gr

发布时间：—

抓取时间：2026-08-20 12:22

来源机构：Hugging Face