Towards Efficient Keyframe Selection for News Video Captioning
Bibliographic record
Abstract
Vision-language models are increasingly used for video captioning, but their reliance on uniform frame sampling introduces inefficiencies. Uniform sampling risks overlooking informative frames rich in textual or structural cues while redundantly processing visually similar frames, leading to inflated computational costs and degraded caption quality. To address this, we propose a layout-aware keyframe selection method that integrates shot segmentation with layout detection to identify frames containing the most semantically informative visual and textual elements. By prioritizing event-rich frames, our approach reduces redundancy while preserving critical contextual information for downstream captioning. On a benchmark of English-language news videos, our method achieved an event matching F1 score of 0.961, a 14% improvement over uniform sampling, while reducing the number of selected frames by 12.78% compared to state-of-the-art methods. These findings underscore the importance of task-specific frame selection in optimizing VLM-based video captioning, particularly for domains rich in on-screen text such as news broadcasts.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.003 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".