Hello everyone,
My question is really hypothetical. I am right now using Cloud vision api with feature web detection. Desc: Detect topical entities such as news, events, or celebrities within the image, and find similar images on the web using the power of Google Image Search.
Does profile login effects the results that web detection will return. For example if I type in google search python, I will get python programming language, while my mum will snake. Does the same logic works for cloud vision API web detection ?
Why this question even raised, because same image executed from 2 different profiles ( I have 2 accounts on cloud Vision API) the model return different probabilities.
Solved! Go to Solution.
Yes, it is indeed an intended behavior getting different results even if you use the same image.
The API returns matching results with their respective scores. The scenario you are experiencing would entail that the scores are relatively close to one another and the API cannot properly select which is a better representation of the image being provided. For more accurate results, we would advise using product reference images with bounding poly coordinates[1].
[1]https://cloud.google.com/vision/product-search/docs/tutorial#5_create_a_products_reference_image