Occupancy Counting Based on IR Arrays

The project uses a neural network to classify human entrance and exist based on the IR arrays from the sensor AMG8833

Contents

• Recording
• Data Cleaning
• Arduino
• Model

Recording

record.py

This script is responsible for recording the training dataset. To use this code, a Raspberry Pi Pico is required to connect the AMG8833 sensor to your local machine. For setup and usage instructions, refer to this detailed tutorial.

Data Cleaning

data_cleaning.ipynb

Jupyter Notebook containing the code for cleaning the dataset and performing data augmentation. This notebook outlines the steps taken to prepare the data for modeling.

random_forest.ipynb

This Jupyter Notebook contains the code for feature extraction from raw data using a random forest algorithm.

Arduino

For programming the Raspberry Pi Pico using the Arduino IDE, please follow the instructions provided by arduino-pico. This will guide you through setting up your development environment.

Inference

This folder contains the Arduino code for inference. It includes the main logic for processing and interpreting sensor data.

Inference_adafruit_mqtt

This directory contains Arduino code for publishing inference results via MQTT using Adafruit IO.

Model

ei-human-entrance-final-arduino-1.0.3.zip

This compressed file contains the machine learning model converted into an Arduino library format.