IBM watson visual recognition services _visualRecognition.Classify() is not returning proper result for Custom classes

0
votes

I am developing VR tool in .NET framework using IBM watson visual recognition service. _visualRecognition. Classify () method was working fine for the custom classifiers before a week. Now I am running the same code but it's not working properly, it's not classifying any images with respect to created custom classes. It's working as default classify method even after passing classifierID's and Owner Id's. It's work as default classify method

Code:

var result = _visualRecognition.Classify(imageByte, imagePath, "image/jpg", Urls, Classifiers, Owners, 0.8f, "en");

Before same code returning below result. Please refer below image:

Result running same code before

Result "One" class in Custom classifiers.

But now same code is returning different result:

Same image, but returning different result

1
ibm-watsonvisual-recognition
You haven't provided a clear statement of the problem. If you've changed the code, you should say so and show the code. If you are working with different data than before, state that. – ShpielMeister
var result = _visualRecognition.Classify(imageByte, imagePath, "image/jpg", Urls, Classifiers, Owners, 0.8f, "en");, This is my code. It was classifying image as per our created custom classifier. I am running the same code after a week but it's not giving right result. – Hanumantapppa Budihal
Not enough information to answer your question. Did you have both classifications before? Have you added new training data? What's the confidence level of the before and after? What's the confidence level of the one not showing before/after? Do you get the same results if you put your classifier in a different domain? – Simon O'Doherty
Ya I trained and last week it's working fine but now I am running the same code, not getting correct result ( I am working on .NET SDK) – Hanumantapppa Budihal

1 Answers

2
votes

We updated the service last Monday and a slight change in how certain options took precedence caused a problem with users of the node SDK. If you update your node SDK from github it should resolve the problem. Sorry for the inconvenience!