Computer Vision News - August 2016

Our CEO Ron Soferman has initiated a series of lectures providing a robust yet simple overview about how to ensure that computer vision projects respect goals, budget and deadlines. Mr. Soferman shares one of his project management tips in each Computer Vision News issue. This month, we will learn about Project Time Estimation . One of the most troublesome questions for project managers is how to correctly estimate project times when research tasks are involved. The tendency is to ask an estimate to development people, whose background is scientific. In our experience, this is a sort of paradox: most scientists, when they see a topic which can be solved, give a tie estimate which is relatively short (in the order of weeks) and which represents as a matter of fact their confidence in their method, in the technologies which they plan to use and in their capability to use them. We are very pleased to receive estimates of this kind: this is proof of the clearness of thought with which the scientist is preparing his or her job. However, it is not so sure that this is a correct estimation of the quantity of work which needs to be done. Actually, the project manager would better prepare an estimation of his own, based on similar projects conducted in the past and with a thorough understanding of the software projects in its entirety. The R&D project estimation must include stages like creating the software environment, the first algorithm version and several iterations until the desired results are reached. Successful projects must include activities like testing, debugging, verification, validation and eventually improving software versions. It is not surprising that in that way a two-weeks project needs in reality a full four months of work. Those two weeks will be used to write the technical basis for the algorithmic work; after which, the project can start its iterations, dealing with real- world problems and with all that does not exactly work as in the textbook. New algorithms and new layers will be then added to the projects in order to drive it to the desired results. This is why we recommend to assess estimates from similar projects, rather than look for internal clues within the project itself. Asking development people how long it will take can lead to an accurate prediction only for short-term tasks, like building a specific algorithm. “ Assess estimates from similar projects, rather than look for internal clues ” 20 Computer Vision News Project Management Tip Project Time Estimation Management