About Object Localization in Vision API

[kuohsin_tu]

Hi,

We are considering using the Object Localization feature in the Vision API, however, we cannot find any information about supported object classes. Is the information open to the public? If Yes, where can we find the information?

Thank you.

[abdelilahf]

Hi,

would you please give more details about the object classes and its meaning? if you give an example with other services will be very helpful. The bellow information is my findings and I need to confirm if that what you have being asking about?

The public documentation [1], mentions only that Object localization identifies multiple objects in an image and provides a LocalizedObjectAnnotation for each object in the image.

The only place where classes are mentioned, for instance:
.net client library, all operations are performed through the following client classes: ImageAnnotatorClient, ProductSearchClient

Java client library, shows the classes supported [3].

[1]https://cloud.google.com/vision/docs/object-localizer
[2]https://cloud.google.com/dotnet/docs/reference/Google.Cloud.Vision.V1/latest/index
[3]https://googleapis.dev/java/google-cloud-vision/latest/index.html

[kuohsin_tu]

Hi, abdelilahf,

Thanks for your detailed information. Sorry for not giving a clear description of my inquiry. I would like to know what kind of objects (for example, bicycles and doors) will be detected in the Object Localization feature of the Vision API. If Google can provide a list of supported objects, that would be helpful.

Thank you.