I have 2 search questions:
I want to do a visual search - find similar images to a given image
I have ~55,000,000 many of them of people and new images are uploaded each day.
I have read some web resources. for example pinterest which I found interesting: https://labs.pinterest.com/user/themes/pinlabs/assets/paper/visual_search_at_pinterest.pdf
- Are there more interesting papers I should read about this topic?
- If I use some model let’s say: ResNet152. On how many of my images should I retrain it (transfer learning)… I assume the more is better.
- How should I do the nearest neighbor retrieval?
Another question about search by color.
For example google have it:
Is it also deep learning? How can it be trained without a color label?
Are there some papers about it?
Thanks in advance,
Anna