I am trying to implement an Image Recommender-System. For this I have two Parts.
- Syntethic created User-Data
- MS-COCO Dataset from 2017 with annotations
I need to create Feature-Vectors from both to combine them in the next step and finally feed them into an MLP to generate Recommendations. I mean for the COCO-Dataset the Feature-Vectors are static because the pictures are not going to change.
My Question in this case is: Is there already created Feature-Vectors from the COCO-Dataset or do I need to generate this myself ? If I need to create this myself, what is the best way to to this ?
Hope you can help me. Thanks for answers
I googled; tried to generate myself some, but quite not sure if this is the right path