I have six million images in amazon s3, and I want to use deep learning models to extract features and run comparisons on those feature vectors to identify duplicates. there are some libs out there that can help with this. I need for this to be implemented using tensorflow.
We have some helper code you can use an aws spot instance with docker container to retrieve images from s3 and process images. This means you only have to focus on the deep learning parts. And even for this, there are some useful libs that we'd like to use like faiss or spotify annoy tree.
Looking to find visually exact images, not similar images. Images may differ in size, file type, for example but can be visually the same.
Will only consider fixed price bids.
AWS experience is a must.