arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2601.14568cs.CVcs.AI
收藏

打破精度-资源困境:一种轻量级自适应视频推理增强

Breaking the accuracy-resource dilemma: a lightweight adaptive video inference enhancement

State Key Laboratory of Radio Frequency Heterogeneous Integration (Shenzhen University)(无线电频率异构集成国家重点实验室(深圳大学));Institute of Applied Artificial Intelligence of the Guangdong–HongKong–Macao Greater Bay(粤港澳大湾区应用人工智能研究院);Henan Academy of Science Applied Physics Institute Co.,Ltd.(河南省应用物理科学研究院有限公司)

Wei Ma, Shaowu Chen, Junjie Ye, Peichang Zhang, Lei Huang

AI总结:

本文提出了一种轻量级自适应视频推理增强框架,通过动态切换不同规模的模型来平衡资源利用与推理性能。

摘要:

现有的视频推理(VI)增强方法通常通过扩大模型规模和采用复杂的网络架构来提高性能。尽管这些方法展示了最先进的性能,但往往忽视了资源效率和推理有效性之间的权衡,导致资源利用效率低下和次优的推理性能。为了解决这个问题,本文开发了一种基于关键系统参数和推理相关指标的模糊控制器(FC-r)。在FC-r的指导下,提出了一种VI增强框架,利用相邻视频帧中目标的时空相关性。根据目标设备的实时资源条件,该框架可以在VI过程中动态切换不同规模的模型。实验结果表明,所提出的方法有效实现了资源利用和推理性能之间的平衡。

英文摘要:

Existing video inference (VI) enhancement methods typically aim to improve performance by scaling up model sizes and employing sophisticated network architectures. While these approaches demonstrated state-of-the-art performance, they often overlooked the trade-off of resource efficiency and inference effectiveness, leading to inefficient resource utilization and suboptimal inference performance. To address this problem, a fuzzy controller (FC-r) is developed based on key system parameters and inference-related metrics. Guided by the FC-r, a VI enhancement framework is proposed, where the spatiotemporal correlation of targets across adjacent video frames is leveraged. Given the real-time resource conditions of the target device, the framework can dynamically switch between models of varying scales during VI. Experimental results demonstrate that the proposed method effectively achieves a balance between resource utilization and inference performance.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑