README.md

Columbia University Image and Face Search package

This package is the key component of the image and face search tool developed for the MEMEX project.

The package is divided into multiple sub-modules:

api: Flask API to expose the face search index.
common: some common resources, like configuration reader, error printing.
detector: detect faces in images.
extractor: generic extraction process defined by a detector and featurizer.
featurizer: compute a discriminative feature, from a face bounding box if provided.
imgio: common methods to download images and get images infos.
indexer: interaction with a database (currently HBase) to get new images and images metadata.
ingester: ingest images from a data source (currently local folder or Kafka).
searcher: search indexing scheme (currently LOPQ).
storer: store data locally or to S3.
updater: check for new images to be processed and process them.

Detailed documentation can be build, see folder docs.