A main design goal of this package is to provide simple separate tools that each solve a simple constrained problem while allowing for interaction between these tools in a loosely coupled way.
This package currently provides:
- A tool that implements an image similarity search on the basis of a neural network model that provides image features.
- A tool to annotate multiple regions in an image with multiple tags.
- A tool to annotate images with on unique label per image.
- A tool that trains an image classifier on the basis of the annotations.
Required python version is 3.8. Consider use of pyenv if that python version is not available on your system.
Activate virtual environment (virtualenv):
source venv/bin/activate
or (pyenv):
pyenv activate my-python-3.8-virtualenv
Update pip:
pip install -U pip
Install sbb_images:
pip install -e ./
All the tools require a SQLITE database that describes the set of images you are working on. That database does not store the actual images but only their path. You can create that image database using a command-line tool provided by this package:
create-database --help
Usage: create-database [OPTIONS] DIRECTORY SQLITE_FILE
DIRECTORY: Recursively enlist all the image files in this directory. Write
the file list into the images table of SQLITE_FILE that is a sqlite3
database file.
Options:
--pattern TEXT File pattern to search for. Default: *.jpg
--follow-symlinks
--subset-json PATH Consider only the subset of image files defined in
this json file.
--subset-dirs-json PATH Recursively search only through a subset of sub-
directories as defined in this json file.
--help Show this message and exit.
In order to speed up things, some command line tools of this package can use a pre-compiled thumbnail database instead of loading the images directly from disk. Depending on the type of disk storage this pre-compilation can heavily improve processing speed.
That thumbnail database can be created follows:
create-thumbnails --help
Usage: create-thumbnails [OPTIONS] DIRECTORY SQLITE_FILE
DIRECTORY: Recursively enlist all the image files in this directory and add
them to the thumbnail database. Write the thumbnail information to the
thumbnails table of SQLITE_FILE that is a sqlite3 database file.
Options:
--pattern TEXT File pattern to search for. Default: *.jpg
--follow-symlinks
--subset-json PATH Consider only the subset of image files defined in
this json file.
--subset-dirs-json PATH Recursively search only through a subset of sub-
directories as defined in this json file.
--max-img-size INTEGER Scale all the images before storing such that the
maximum of their width and height is equal to this
values (default 250).
--processes INTEGER Number of parallel processes to be used.
--help Show this message and exit.
Note that your current pwd should be from where you want to run the webapp (annotator, region-annotator or image search) later on.