Search papers, labs, and topics across Lattice.
This paper introduces MetaVideoAgent, a framework that automates the evolution of video agents tailored for long-form video understanding by profiling information density and evidence requirements from sampled frames. By compressing localized failures into minimal validation tasks and employing a modular agent representation, the framework effectively updates only the necessary components, leading to significant improvements in accuracy. The results show that MetaVideoAgent enhances macro-average accuracy from 38.44% to 51.47% over four evolution iterations, outperforming existing fixed-design agents while minimizing resource usage.
Evolving video agents can boost long-form video understanding accuracy by over 13 percentage points while dramatically reducing processing costs.
Long-form video understanding requires locating sparse, question-relevant evidence in long, multimodal videos. Real-world video distributions differ in modality-specific information density, content structure, and evidence patterns, causing fixed video-agent designs to incur redundant processing or fail when mismatched. Extending automated agent evolution from text to video is challenging because full long-video execution makes candidate validation expensive, failures propagate across coupled evidence-processing stages, and complex preprocessing, perception tools, and localization strategies make code-level updates difficult to implement reliably. We introduce MetaVideoAgent, a framework that automatically evolves a video agent for a target distribution. It profiles information density and evidence requirements from sparsely sampled frames and associated queries to guide initial design, then compresses localized failures into independently executable minimal validation tasks. It constructs evidence-grounded Gold Paths, audits Student trajectories, aggregates recurring failures across samples, and attributes them to responsible modules. A modular agent representation constrains each update to the primary responsible module and its necessary dependencies. We further introduce VA-EvoBench, covering eight video distributions with separate evolution and held-out splits. With four evolution iterations per distribution, MetaVideoAgent improves every initial agent and raises macro-average accuracy from 38.44% to 51.47%, at an average evolution cost of 3.54M tokens per distribution. The evolved agents outperform the strongest prior fixed-design video agent by 6.39 percentage points while using the fewest tokens and video frames per question among the compared video agents. We will release all code and data to support reproducible research.
