Project Voldemort

This repo exists to summon You-Know-Who, He-Who-Shall-Not-Be-Named, otherwise known as the Dark Lord.

It does not do anything useful.

Pre-Requisites

• Obtain a facebox from machinebox.io
• Make sure the facebox is running
• .env must set MB_KEY
• docker-compose start starts dependencies

Data Processing Pipeline

• Gather images of the person you would like to detect
• Run voldemort data to generate the training data in training/data.json
• Run voldemort train to teach the facebox how to find Tom Riddle.

Face Detection

• Run voldemort identify --image <URL> to process the image
• Response JSON will contain matches

Example Response:

{
	"success": true,
	"facesCount": 2,
	"faces": [
		{
			"rect": {
				"top": 573,
				"left": 1294,
				"width": 186,
				"height": 185
			},
			"matched": false
		},
		{
			"rect": {
				"top": 201,
				"left": 386,
				"width": 186,
				"height": 186
			},
			"id": "59051e68e570abb39ece3c6395d72282",
			"name": "Tom Riddle",
			"matched": true
		}
	]
}

Initialize Facebox State

• .state file can be used to initialize MB_FACEBOX_STATE_URL