HORNet: Task-Guided Frame Selection for Video Question Answering with Vision-Language Models
This addresses efficiency and accuracy issues in video question answering for researchers and practitioners using vision-language models, though it is incremental as it builds on existing frame selection methods.
The paper tackled the problem of inefficient frame selection in video question answering by introducing HORNet, a lightweight policy that reduces input frames by up to 99% and processing time by up to 93%, while improving answer quality with gains like +1.7% F1 on MSVD-QA and +7.3 points on NExT-QA.
Video question answering (VQA) with vision-language models (VLMs) depends critically on which frames are selected from the input video, yet most systems rely on uniform or heuristic sampling that cannot be optimized for downstream answering quality. We introduce \textbf{HORNet}, a lightweight frame selection policy trained with Group Relative Policy Optimization (GRPO) to learn which frames a frozen VLM needs to answer questions correctly. With fewer than 1M trainable parameters, HORNet reduces input frames by up to 99\% and VLM processing time by up to 93\%, while improving answer quality on short-form benchmarks (+1.7\% F1 on MSVD-QA) and achieving strong performance on temporal reasoning tasks (+7.3 points over uniform sampling on NExT-QA). We formalize this as Select Any Frames (SAF), a task that decouples visual input curation from VLM reasoning, and show that GRPO-trained selection generalizes better out-of-distribution than supervised and PPO alternatives. HORNet's policy further transfers across VLM answerers without retraining, yielding an additional 8.5\% relative gain when paired with a stronger model. Evaluated across six benchmarks spanning 341,877 QA pairs and 114.2 hours of video, our results demonstrate that optimizing \emph{what} a VLM sees is a practical and complementary alternative to optimizing what it generates while improving efficiency. Code is available at https://github.com/ostadabbas/HORNet.