recops developed on x64 Linux box using Python version 3.10.2. Is expected to work flawlessly on every linux box with Python versions above 3.7. In order to run on other systems we suggest to use the Dockerized version.
First of all we clone recops repository:
git clone https://github.com/forensic-toolkit/recops/
cd recops
In order to install recops we will create a python virtual environment, setup required libraries, and download the pre-trained models.
Create virtual environment for python:
# Create a directory
mkdir -p ~/.recops/
# Create python virtual environment
python -m venv ~/.recops/venv
# Activate python virtual environment
source ~/.venv/bin/activate
# Upgrade pip
pip install --upgrade pip
Install requirements and recops:
# Install recops requirements
pip install -r requirements.txt
# Install recops
python setup.py install
Download the pre-trained models:
recops-download-models.sh
In order to install recops we will build a docker image, and run it:
# Build docker image
docker build -t recops .
# Run docker image
docker run -it recops --help
By default the docker image stores its data under /var/lib/recops inside the container. To persist the data use the -v option:
docker run -v /full/path/to/perstist:/var/lib/recops -it recops --help
If you want to get a bash shell to your container you can change the entrypoint:
docker run --entrypoint /bin/bash -it recops
Last but not least it might be helpful to run the docker container in the background using the webui command and execute recops at the running container like so:
docker run \
--name my-recops \
--volume /tmp/my-data:/var/lib/recops \
--publish 127.0.0.1:5000:5000 \
--detach recops webui --host 0.0.0.0
# Open web-ui on our browser
firefox http://127.0.0.1:5000
# Execute recops command on running container
docker exec -it my-recops recops --help
Following is the default list of filepaths where recops stores data.
~/.recops/venv/.. => Default location of python virtualenv (see above for installation)
~/.recops/data => Default location where recops stores its data.
~/.recops/data/recops.db => Default Sqlite DB location.
~/.deepface/weights => Deepface library uses this location to store pre-trained models.
In order to change the location of the database and files you can specify following environment variables:
export STORAGE_URI=file:///run/media/my-external-hard-disk/data
export DATABASE_URI=sqlite:////run/media/my-external-hard-disk/data/recops.db
When your dataset contains sensitive data you might want to move ~/.recops/data to an encrypted LUKS container or another location of an external hard drive. See this simple guide on how to do so.
Usage: recops [OPTIONS] COMMAND [ARGS] ...
Options:
--log-level TEXT Specify logging level
--log-file TEXT Specify file to output logs (by default stderr)
--help Show this message and exit.
Commands:
backup Backup full database and files.
dataset-cluster-faces Cluster faces for given dataset
dataset-compute-extended-fields Use extended models to compute...
dataset-create Create dataset
dataset-delete Delete specified dataset and all...
dataset-export Export datasets content into a zip...
dataset-import-faces Import faces from local folder
dataset-import-identities Import faces linked to identities from...
dataset-import-images Import images from local folder
dataset-link-matched-faces Link faces to identities from given...
dataset-list List available datasets
dataset-list-faces List available faces in dataset
dataset-list-identities List available identities in dataset
dataset-list-images List available images in dataset
dataset-match-faces Compare faces linked to identity with...
debug
face-delete Delete a face
identity-create Create an identity
identity-delete Delete an identity
identity-link Link existing identity to a face
image-delete Delete an image
recognize Runs face recognition on a given image...
verify Runs face verification on a given...
webui
NOTE: recops is still under development and code is subject to major changes. Please keep in mind that the release version 0.1.5 is a BETA version.
This document here thoroughly explains recops internals, design and functionality, we suggest reading it to get a deeper understanding.