Images tagged using Computer Vision and uploaded to Custom Vision via python SDK are not showing the tags.

SANJEEV ASHER 0 Reputation points
2024-04-11T08:20:26.04+00:00

I have a bunch of images with Vehicles and People on the road. I have used Computer Vision API to detect the objects and get the tags and coordinates for each object in the images. I used the same images along with the regions and tags and uploaded them to Custom Vision via API. All the images are getting uploaded, but none of the tags are visible. Can you please help identify the issue. I have created the exact tags as in the images in Custom Vision and also normalized the coordinates for the objects.

Following is my debug log for detecting the objects via Computer vision

Objects in image: <_io.BufferedReader name='IMG_5811.jpg'>

 person

 motorcycle

Objects in image: <_io.BufferedReader name='IMG_5805.jpg'>

 Land vehicle

 truck

Objects in image: <_io.BufferedReader name='IMG_5839.jpg'>

 car

 car

 motorcycle

 person

 motorcycle  

Following is the debug log for the image upload to Custom Vision
IMG_5811.jpg person 0.425 0.4025 0.25666666666666665 0.4575

IMG_5811.jpg motorcycle 0.36333333333333334 0.55625 0.5016666666666667 0.3925

Uploaded image IMG_5811 with 2 regions.

IMG_5805.jpg Land vehicle 0.91 0.27875 0.09 0.3425

IMG_5805.jpg truck 0.24833333333333332 0.235 0.7083333333333334 0.61375

Uploaded image IMG_5805 with 2 regions.

IMG_5839.jpg car 0.7466666666666667 0.43875 0.16333333333333333 0.0825

IMG_5839.jpg car 0.935 0.43875 0.065 0.08375

IMG_5839.jpg motorcycle 0.06666666666666667 0.44 0.23166666666666666 0.125

IMG_5839.jpg person 0.3433333333333333 0.37 0.2733333333333333 0.43

IMG_5839.jpg motorcycle 0.27166666666666667 0.4625 0.5466666666666666 0.4925

Uploaded image IMG_5839 with 5 regions.

Azure Computer Vision
Azure Computer Vision
An Azure artificial intelligence service that analyzes content in images and video.
371 questions
Azure AI Custom Vision
Azure AI Custom Vision
An Azure artificial intelligence service and end-to-end platform for applying computer vision to specific domains.
247 questions
{count} votes

Your answer

Answers can be marked as Accepted Answers by the question author, which helps users to know the answer solved the author's problem.