Team solutions

Turtles – Case Kaufland

In this article we present our solution for helping customers and making their shopping experience easier while identifying products from images. We bring forward our idea and discuss the results of our CV experiment.

2
votes

2 thoughts on “Turtles – Case Kaufland

  1. 0
    votes

    You show good understanding of the data set and the challenges that it represents.
    You’ve managed to find an interesting workaround some of the issues mainly orientation and lighting.
    However there are some areas that you can improve in. First is the business understanding. You need to show that you have understood the problem domain and who the final users of the system will be. What their expectations are? In the model section you make a very good argument as to why you’ve decided to use GoogleLenet, however you could go into more details about the advantages and disadvantages of using this tech as opposed to other solutions. In the evaluation section you need to provide an analysis of the results showing that you understand the inner workings of your system, not just raw results.
    Also you should never provide the results from the training and validation sets
    as those are considered training data and the results on them are misleadingly high, due to the fact that the system has seen them before. You should always have a separate set of data for testing purposes.

Leave a Reply