You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is your feature request related to a problem? Please describe.
Be able to group similar images/videos together. Users can then decide whether to keep all or delete some.
Next step: highlight the best from the group. (It is not blurry, and there is good lighting, etc.)
Describe the solution you'd like
Use CNN models like ResNet and VGG19 to create embeddings
Compare embeddings using the Cosine Similarity algorithm to group similar images
Use an image aesthetic scoring algorithm to find the best-looking image
Additional context
Can make this more fast and efficient using a vector database
The text was updated successfully, but these errors were encountered:
Currently saves the vector embeddings from each image in a .pt file. This file is very large and expensive to load every time the server needs to find similarities. A vector database might help with this problem.
Is your feature request related to a problem? Please describe.
Be able to group similar images/videos together. Users can then decide whether to keep all or delete some.
Next step: highlight the best from the group. (It is not blurry, and there is good lighting, etc.)
Describe the solution you'd like
Additional context
The text was updated successfully, but these errors were encountered: