Computer Vision News - June 2016

Our 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. Mr. Soferman shares one of his project management tips in each Computer Vision News issue. This month, we will learn about other aspects of input data diversity ; in particular, the influence that input variety has on product requirements and functionality. This project required the preparation of algorithms for a flower grading machine , programmed to provide the grading of each flower (see Computer Vision News of April 2016 ) . In this project, the challenge stems primarily from the difficulty to predict the kinds of flowers which will need to be sorted and therefore the resources which must be allocated to such a project. Our decision in that regard was to start with one specific flower : we developed the algorithm, which was very successful. Then we moved to another flower, then to the next one. All three responded successfully to what we were looking for: the flower grower was obviously very satisfied, since that enabled him to display his produce at the Amsterdam flower auctions with the high precision of a sophisticated and objective third party solution. Unfortunately, after the first three kinds of flowers, things got complicate. The complexity of the projects is often due to the complexity of nature: the view of the stems might be difficult since they may be covered by leaves, hindering the correct assessment of the stems width, one of the key features in the flower market as it is considered a reliable indicator of their expected life span . The differing stock exchange requirements for each flower (in particular regarding the packaging in bins) also interfere, so that this project cannot be a general one: it will always have to be defined in relation to a specific flower. “ The complexity of projects is often due to the complexity of nature ” The best example to illustrate the influence of input data diversity is a flower sorting and grading project we have done in the past. 26 Computer Vision News Project Management Tip More on Data Diversity