Abstract: In video analytics, accuracy and efficiency are two important metrics and there tend to be a tradeoff between each other. In this paper, we consider accuracy-efficiency optimization for ...
To addressing the challenges of unsupervised and non-differentiable sparse frame sampling in Video-MLLMs, We propose Temporal Sampling Policy Optimization (TSPO), a reinforcement learning framework ...
Abstract: The streaming of 360-degree video faces significant bandwidth shortage due to its ultra-high resolution. To address bandwidth constraints, viewport prediction-based methods prioritize ...
Have you ever wondered if the file format of your video can automatically determine the best frames to pass to a multimodal large language model? Have you ever wondered if this approach can give the ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results