Human readable description with Amazon Rekognition

Amazon Rekognition can bring you the tags, which are the objects that are recognized in the image.

Using the combination of services you propose will not help you solve the problem either. You should explore other alternatives.

I recommend you check the following post where they tell you about the Image-to-Text problem

https://huggingface.co/tasks/image-to-text

I hope I've helped.