# dKFD: Phase-Structured Evidence Allocation for Fixed-Budget Localized Event Understanding

arXiv: https://arxiv.org/abs/2609.33083 · HTML: https://mlsanity.com/paper/2609.33083/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 21 of its 30 references that are in the map.
- First posted on arXiv 27 Sep 2026.

## Builds on (most cited first, 21 of 21)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Qwen2.5-VL Technical Report](https://mlsanity.com/paper/2502.13923/) (2025)
- [Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset](https://mlsanity.com/paper/1705.07750/) (2017)
- [LLaVA-OneVision: Easy Visual Task Transfer](https://mlsanity.com/paper/2408.03326/) (2024)
- [SlowFast Networks for Video Recognition](https://mlsanity.com/paper/1812.03982/) (2018)
- [InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models](https://mlsanity.com/paper/2504.10479/) (2025)
- [Temporal Segment Networks: Towards Good Practices for Deep Action Recognition](https://mlsanity.com/paper/1608.00859/) (2016)
- [ActionFormer: Localizing Moments of Actions with Transformers](https://mlsanity.com/paper/2202.07925/) (2022)
- [Learning 2D Temporal Adjacent Networks for Moment Localization with Natural Language](https://mlsanity.com/paper/1912.03590/) (2019)
- [UniVTG: Towards Unified Video-Language Temporal Grounding](https://mlsanity.com/paper/2307.16715/) (2023)
- [Streaming Long Video Understanding with Large Language Models](https://mlsanity.com/paper/2405.16009/) (2024)
- [Adaptive Keyframe Sampling for Long Video Understanding](https://mlsanity.com/paper/2502.21271/) (2025)
- [Query-Dependent Video Representation for Moment Retrieval and Highlight Detection](https://mlsanity.com/paper/2303.13874/) (2023)
- [TriDet: Temporal Action Detection with Relative Boundary Modeling](https://mlsanity.com/paper/2303.07347/) (2023)
- [CLIP-It! Language-Guided Video Summarization](https://mlsanity.com/paper/2107.00650/) (2021)
- [Interventional Video Grounding with Dual Contrastive Learning](https://mlsanity.com/paper/2106.11013/) (2021)
- [VideoLLM-MoD: Efficient Video-Language Streaming with Mixture-of-Depths Vision Computation](https://mlsanity.com/paper/2408.16730/) (2024)
- [BIMBA: Selective-Scan Compression for Long-Range Video Question Answering](https://mlsanity.com/paper/2503.09590/) (2025)
- [VRU-Accident: A Vision-Language Benchmark for Video Question Answering and Dense Captioning for Accident Scene Understanding](https://mlsanity.com/paper/2507.09815/) (2025)
- [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://mlsanity.com/paper/2303.04995/) (2023)
- [SEAL: Semantic Attention Learning for Long Video Representation](https://mlsanity.com/paper/2412.01798/) (2024)

## Abstract (arXiv)

Sparse video understanding often requires selecting a small set of visual evidence under a fixed frame budget. Most sparse selectors allocate this budget globally, allowing all frames to compete with one another. For temporally localized events, this can be a poor inductive bias: useful evidence is often distributed across pre-event context, the event itself, and post-event consequences. We study fixed-budget evidence allocation for localized event videos and show that globally competitive Top-K selectors can preserve recognition and grounding while producing unstable event evidence. On DoTA Video Anomaly Recognition, Global Top-K obtains competitive recognition and temporal grounding, but low selector-event alignment at K=12 (Frame AUC 51.5 ± 10.8). We propose dKFD, a phase-structured differentiable selector that reserves evidence capacity across pre-event, event, and post-event phases after full-sequence temporal encoding. Under matched-budget multi-seed evaluation, dKFD improves Frame AUC by +30.97 over a matched Global Top-K selector at K=12 (p<0.01), while yielding modest but statistically significant recognition gains and comparable temporal grounding. Mechanism ablations show that phase supervision is load-bearing: removing it reduces Frame AUC to 41.1 ± 12.0 even when phase-partitioned budgets are retained. Downstream diagnostics on VRU-Accident show consistent gains over learned Global Top-K across VLM families, while dense captioning reveals a boundary condition where uniform sampling remains competitive. These results support phase-structured allocation as a controlled fixed-budget approach for event-centric sparse evidence selection, not as a universal video summarization strategy.