- Rating
- 2
- Confidence
- 2
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
Multi-video reasoning here is treated as a sequence of decisions about which video and temporal segment to inspect before answering. VideoInspector-Instruct supplies approximately 95K compositional examples covering retrieval, temporal grounding, clip reordering, and comparison, with synthetic tool-use and error-recovery trajectories. VideoInspector learns from these examples through supervised fine-tuning followed by agentic reinforcement learning. Evaluation on CrossVid, CVBench, and VEBench shows improvements over both backbone models and their trained variants without tools.
Critical strengths
-
The gains extend across both backbones and all three benchmark summaries. CrossVid accuracy improves from 18.7 to 29.1 on the weaker backbone (Table 2), supporting the practical value of the supervision.
-
The four training tasks cover complementary capabilities: retrieval, localization, ordering, and comparison. Removing any task reduces performance in Table 3, supporting the combined training design.
Critical weaknesses
-
I am not convinced the gains establish better evidence acquisition and self-correction. The comparison without tools changes both supervision and access to segment observations; the stage ablation does not disentangle these effects. A comparison with fixed inspections under matched training and visual budgets would help. VEBench evaluates final localization, but intermediate evidence support and recovery remain untested quantitatively.
-
Dataset validation remains too limited for a central contribution. Section 3.2 describes automatic filtering and manual spot checks, but reports neither the audit size nor measured annotation error rates. The number of unique source videos and benchmark overlap checks are also unspecified. A quantitative human audit and documented overlap checks would strengthen confidence in the supervision and reported transfer.
-
Inference efficiency is not reported. Section 5.1 lists hardware and training settings, but gives no latency, average tool-call count, or inference cost. Repeated segment inspection could make the accuracy gains expensive. Without an accuracy–cost comparison against the baselines, I cannot assess whether the improvement justifies the additional computation.
Questions for authors
-
What exactly does
inspect_segmentreturn during training and evaluation: sampled frames, generated descriptions, or both? What are the initial sampling settings and call limits, and how are visual budgets controlled across comparisons? -
Under matched training and visual budgets, how does learned inspection compare with fixed or uniformly selected inspections? Can you also report recovery success after an incorrect initial inspection, with and without recovery-trajectory supervision? These results could change my recommendation.
-
How many unique videos underlie the dataset, what quantitative results were obtained from manual validation, and what checks excluded overlap with CrossVid, CVBench, and VEBench?
-
On identical video–question inputs and comparable hardware, how do end-to-end latency, tool-call counts, and inference cost compare with the variant without tools and the strongest agentic baselines? Do the accuracy gains persist under the same time or compute budget?