ACAME: Advanced Control Aspects for MEdia workload processing

Dithen是一个计算即服务平台，专为并行执行大规模多媒体任务而设计。通过使用Dithen，企业或最终用户可以在大图像/视频数据集中进行人脸或物体的检测和识别、视觉相似性识别、视频转码等，而无需管理任何云计算服务器，只需通过非常简单直观的web服务即可。通过这种方式，Dithen的用户无需管理其软件或硬件基础设施，也无需将本地多媒体处理软件转换为适合云计算的框架，就可以轻松拥有一台超级计算机。在这个项目中，我们建议研究和开发来自控制系统理论的先进概念，以优化Dithen的部署，从而将其在云计算基础设施上的支出减少一个数量级。

Dithen is a Computation-as-a-Service platform tailored to the parallel execution of very-large scale multimedia tasks. By using Dithen, businesses or end users can carry out face or object detection and recognition, visual similarity identification, video transcoding, etc., within big image/video datasets without managing any cloud-computing servers, but simply via a very simple and intuitive web service. In this way, users of Dithen have a supercomputer at their fingertips without having to manage its software or hardware infrastructure, or indeed having to convert their local multimedia processing software to a framework amenable to cloud computing. In this project we propose to research and develop advanced concepts from control systems theory for the optimization of the deployment of Dithen, in order to decrease its expenditure on cloud-computing infrastructure by up to an order of magnitude.