This software is a Python Desktop Application and Server that uses Tensorflow, Redis, FFMPEG and such libraries to detect and recognize faces in video streams and photos.
- ADD model downloads to
install.sh - COLLECT all model files to a
modelsdirectory ignored by git. - CREATE a server python file that runs only detection and recognition modules
install.sh: Install necessary packages and libraries on a Ubuntu 18.04 installation and creates virtual environment.
facebin.sh: Runs the desktop application by activating the environment and running python command. Keeps logs in FACEBIN_DIR/logs/$(date)
facebin_gui.py: Creates the gui and runs the servers. Main entrance of the software.
camera_controller.py: A class to manage cameras and start/stop them using libav
camera_dialog.py: A Qt dialog to control up to 4 cameras
camera_reader.py: Uses libav to read frames from cameras and puts them to Redis server in localhost
database_api.py: Provides database access functions
dataset_manager_v3.py: Provides training dataset access functions.
export.py: Exports the dataset to image files to ~/facebin-artifacts/export-$(time.time)
face_detection.py: Contains Tensorflow and Haar based classes for face detection. Receives image files from Redis queue, runs face detection on them and stores there again.
facebin.db: SQLite database that contains basic information about recognized persons and their images. Also some configuration is done here.
face_recognition_v6.py: Receives faces from Redis, recognizes them using Tensorflow model and puts their information to redis
redis_queue_utils.py: Higher level functions to manage image queues in Redis.
requirements-gpu.txt: Python pip requirements file for GPU based recognition libraries
requirements.txt: Python pip requirements file for CPU based recognition libraries
show_image_dialog.py: Simple QT dialog to show Numpy or QImage based images
history_dialog.py: Shows timestamp, person, image information for the recognized faces. It retrieves these from the database.
history_recorder.py: One of the elements of server processes. Retrieves the recognized faces from Redis and records them to database.
import.py: Imports the initial dataset from a directory by creating necessary DatasetManager object and importing the information.
utils.py: various utilities
person_dialog.py: Shows information about a particular person.
video_recorder.py: Records footage videos to a directory
visualization_utils_color.py: To draw on images
watch-image-dir.sh: Watches the changes to a directory and sends the new images using email
facebin_init.py: It adds CUDA paths to $PATH and imports tensorflow afterwards.