Computer Vision News - August 2018

If you want to move into the deep learning area and manage your project in an optimal way, it is key that you choose the system that best suits your work. A further analysis of this questions finds 3 parameters which must be take into account: a) the number of engineers involved in the project; b) the amount of calculations and data which will be needed for the task; c) the continuity of the project. Once these parameters are known, they enable the project manager to define the right platform for the Deep Learning system. Let’s see them one by one. When the group working on deep learning sends frequent tasks for training, chances are that the computer ends up working non-stop. Given the costs of purchase of a Deep Learning server versus the cost of using a server on the cloud (like on Amazon Web Services or similar services), the former might become more convenient. In that case, the recommendation is to buy one or more servers with strong GPU for the project. On the other hand, if it’s a one-time run, especially when it needs a lot of computation, large datasets and much memory, the best option is to use the cloud services of AWS and the like. In that case, sample data will be run on the local machine and the cloud resources will be used only for the big run. Finally, it might be more convenient to buy your team a laptop with better GPU capabilities. That will enable to start the development with its initial tasks on a local and portable environment. Further major tasks with long hours of GPU work will be done on the other available servers. I hope that these advises help the project manager to scope the resources needed for the Deep Learning project. Deep Learning and GPUs for dummies Management Project Management Tip 21 Computer Vision News RSIP Vision ’s CEO Ron Soferman has launched a series of lectures to provide a robust yet simple overview of how to ensure that computer vision projects respect goals, budget and deadlines. This month we learn about the basic choices for Deep Learning and GPUs . It’s another tip by RSIP Vision for Project Management in Computer Vision . …start the development with its initial tasks on a local and portable environment. Once these parameters are known, they enable the project manager to define the right platform for the Deep Learning system.