Im working on a College project using DeepFace, and one of the requirements is to run a appliciation hosted on the cloud. Is there a way to use DeepFace.find() function setting the db_path as an S3 bucket, or even a table on a database containg the different embeding of my refference images ?

I want to feed the deepface.find a stream of my monitor and and a cloud_based data base ie s3 Bucket or Snowflake table.

Is that something thats possible with Deeface.find() ?

Im working on a College project using DeepFace, and one of the requirements is to run a appliciation hosted on the cloud. Is there a way to use DeepFace.find() function setting the db_path as an S3 bucket, or even a table on a database containg the different embeding of my refference images ?

I want to feed the deepface.find a stream of my monitor and and a cloud_based data base ie s3 Bucket or Snowflake table.

Is that something thats possible with Deeface.find() ?

• python
• deepface

• Please provide enough code so others can better understand or reproduce the problem. – Community Bot Commented Mar 4 at 22:36

1 Answer 1

Find functionality of deepface is only running for files in the file system. If you want to use deepface with a database, then you should use its represent functionality to represent facial images as vector embeddings and then store those embeddings into your database.

If you follow the `Large Scale Facial Recognition` topic in the readme of the repository, it redirects you to several implementation tutorials for relational databases such as postgres or sqlite; nosql databases such as mongo, cassandra, redis; or vector indexes such as annoy, faiss, voyager, elasticsearch; or vector databases such as postgres with pg-vector and redisearch.

You may ask that which one should you use? Relational database, nosql database, vector index or vector database. The number of items in your database will answer this question. If your database size is millionish, then use vector index or vector database. If your database size is ten thousandish, then you can use relational database. If your task requires face verification instead of face recognition, then you should adopt a key value store.