I was looking at object classification and identification services of multiple cloud providers AWS, Microsoft Azure and Google cloud.
I have picked and used the same image and see what would happen (image from https://ankyrahome.com/)
Let's try Amazon
Amazon Rekognition is the service that can help to classify objects and recognize some common features like colors and shapes.
Well it is quite confident (97%) that this is a Rug.. :)
Let's try Microsoft's computer vision and see the results for the same image.
Ok well it is not as confident as AWS rekognition but same result it is still identifying it as a rug.
Now let's try Google vision API and see the result for the same image.
Now this is pretty impressive and super accurate even it has a confidence level of 78%.
As a result, Google vision API is the most accurate one given that there was no additional training and using default demo model.