SParky

Overview

A smart alarm clock supporting configurable timed facial recognition, ambient climate sensing and data querying. Skip to Run to see how to run the application.

Architecture

The architecture of Sparky uses multithreading for concurrent execution of MQTT requests, alarm time checking and ambient climate data upload.

Register/Delete User

Sparky captures multiple facial images and trains a local OpenCV model with those images. The resultant encodings file is stored in the database.

Deleting a user is a simple database call. See register_user and delete_user for its request schema.

Set Alarm

Setting an alarm entails:

• Verifying the user.
• Retrieving their facial encodings from the database.
• Persisting the returned file locally.
• Updating local state to configure the alarm time.

See set alarm for its request schema.

Ring Alarm

Query Data

Specifications

• Facial Recognition: OpenCV
• Database: Azure MySQL
• Message Queue: EMQ MQTT

Requirements

Hardware

• Raspberry Pi (guaranteed to work on a 4)
• Audio output compatible with a Pi.
• Camera compatible with a Pi.
• DHT 11 temperature sensor

Software

• Access to MQTT X Web

Environment Variable file

By design, connecting to a MySQL database or MQTT broker requires a configured .env file. To that end, please follow the steps below:

• Populate the fields without values in template.txt.
• Rename from template.txt to .env.

NB: EMQ provides a free broker. See here for more details.

SQL Tables

The SQL code for generating the data model above is available in tables.sql.

Run

• Install all external dependencies on the import listing.
• If you do not have DHT11 connected, then comment out t_and_h_upload_worker.start() in the main function.
• Execute the main.py file
• Verify database connection was successful through stdout output: Connection established Successfully!
• Go to MQTT X Web and set up a connection to your broker or the free broker.
• Subscribe to output_topic defined in main.py
• Input requests to a topic in input_topics defined in main.py

Requests Schema

Register/Delete User

    username   

• username is an alphanumeric string.

Set Alarm

    username, alarmTime, facialRecognitionDuration    

• username is a registered username.
• alarmTime is a 12-hour format string following HH:MM (A|P)M
• facialRecognitionDuration is the number of seconds the system will maintain continuous facial recognition. The system restarts the timer whenever the recognition fails. Note this value must be lesser than max_facial_recognition_duration which the system uses for timing out the recognition.

Retrieve Data

Temperature & Humidity

    t&h, param   

• param can be either a positive integer specifying number of days before current date or a specific date formatted as mm/dd/yyyy

Alarm data

    username, (columns), param   

• username is a registered username

• param as defined here

• columns is a subset of [wake_up_duration, completed_face_recognition] delimited by a semicolon. If empty, the request becomes username, param

• Example queries:

  username, (wake_up_duration; completed_face_recognition), param   
  

  username, (wake_up_duration), param   
  

  username, (completed_face_recognition), param