Computer Vision News - March 2019

10 Computer Vision News Project Management Tip Management Besides scientific and algorithmic challenges, computer vision work presents also project management challenges: we are going to discuss in this article the specific circumstance of projects that do not have well-defined quantitative goals. Since this is a research project, neither client nor supplier can predict results a priori . One of the first tasks which needs to be dealt with in every project is the need to define our goals, such as would be accepted and understood by both parts, the client and RSIP Vision . In projects of this kind, customers start with some idea of what they want to achieve, but for lack of knowledge of what is possible to achieve, they often expect the service provider to supply the specific operative goals of the project: in other word, how to measure achievements and success at the end of the work. When the objectives are clear and understood, it makes it easier to verify whether we are getting closer or not. When this does not happen, the lack of operative goals has an effect on the whole process. For instance, if you define clearly the goals that you want, you can define the constraints that are upon your work. Some of them are straightforward and some are not: time, amount and quality of data needed. Ideally, the goals of a deep learning project in computer vision should include the level of accuracy that we want to achieve. Sometimes, at the time of the POC , the client has no clear idea of the desired accuracy. That’s equivalent to saying: “ Do the best you can (with the given constraints)! ”. That absence of ground truth might lead to overshooting, i. e. working more than necessary and spend more time than necessary to further improve the results. Without it, gaps may be found in the design: for instance, we have the deep neural network, but how do we translate the output of the neural network to exactly what we need? When you have little time for experimenting, you cannot try everything: you need instead to focus your effort . You define an end-to-end procedure and can admit some errors in the procedure, if the final result is accurate enough. Again, the lack of ground truth denies the chance to evaluate the quality of the result. With When the client says: “Do the best you can!” 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 Aliza Minkov tells us what to do when the client says: “Do the best you can!” . It’s another tip by RSIP Vision for Project Management in Computer Vision . “…a good way of knowing if you are in the good direction…”