predictor.classify_image failed with error "Invalid iteration"

Question

I've created and trained an image classification model using Azure Custom Vision (Cognitive Services) and published the model with API. Now, I've written a simple code in Python which takes an image from my local file and calls the API. However, I'm still getting this error even though the Iteration surely exists:

azure.cognitiveservices.vision.customvision.prediction.models._models_py3.CustomVisionErrorException: Invalid iteration

My code:

ENDPOINT = "https://northeurope.api.cognitive.microsoft.com/"
PROJECT_ID = "my project id"
prediction_key = "my prediction key"

prediction_credentials = ApiKeyCredentials(in_headers={"Prediction-key": prediction_key})
predictor = CustomVisionPredictionClient(ENDPOINT, prediction_credentials)

image_location = "/tmp/123.jpg"
publish_iteration_name = "Iteration2"

with open(image_location, "rb") as image_contents:
    results = predictor.classify_image(
        PROJECT_ID, publish_iteration_name, image_contents.read())

Answer 1

The classificationType shows null, meaning that you created an object detection model. Please follow this example instead to test your model. It should be predictor.detect_image() instead of predictor.classify_image(). Hope this helps!

# Open the sample image and get back the prediction results.  
with open(base_image_location + "images/Test/test_od_image.jpg", mode="rb") as test_data:  
    results = predictor.detect_image(project.id, publish_iteration_name, test_data)  

Answer 2

Hi, the error has to do with the publish name. Ensure that you are entering a valid publish name that you specified when publishing the model. You can retrieve the publish name using the GetIterations API. If you're following this example, the publish name would be publish_iteration_name = "classifyModel". Hope this helps.

Answer 3

I had a similar issue with Microsoft Learn training. Later I found that I did not use the correct model name as in the tutorial, when publishing the model. Once the model name matched the one in the .env python file, the code ran smoothly.