object-based-positioning v1.0.1

Object Based Positioning Library

!NOTE Summary: Prototype mechanism for achieving user positioning in indoor spaces using object recognition and detection models. This library works exclusively with React Native and Expo. A concrete implementation using Firebase to store data is presented, although other mechanisms could be used.

🌟 Introduction

This library is presented as a concrete implementation aimed at enabling a proof of concept. The library allows for positioning users in indoor spaces using various object recognition models. To achieve positioning, there are two stages involved. In the first stage, the user needs to use their phone's camera to scan the environment, recording different objects present in the space. These "points of interest" can then be accessed in the second stage, providing contextual information or services to other users.

🔑 Key Features

🟢 Real-time object detection and classification. 📍 Indoor positioning using object recognition models. ⚙️ Easy integration with React Native and Expo. 🔧 Configurable detection settings and model options.

📚 Table of Contents

1. 🌟 Introduction
2. 🔑 Key Features
3. 📖 Documentation
   • 🔧 BaseObjectBasedPositioning class
   • 🛠️ FirebaseObjectBasedPositioning class

📖 Documentation

BaseObjectBasedPositioning class

FirebaseObjectBasedPositioning class

FirebaseObjectBasedPositioning is a class that extends the functionality of BaseObjectBasedPositioning to provide an indoor positioning mechanism based on storing and managing data in Firebase. This class allows you to register and unregister objects in a Firestore database, retrieve registered objects, and get nearby objects based on the user's current position.

Create an instance

import { Firestore } from 'firebase/firestore';
import { FirebaseObjectBasedPositioning } from './path/to/FirebaseObjectBasedPositioning';

// Initialize Firestore (make sure to replace with your own configuration)
const firestore: Firestore = ...; // Your Firestore initialization here

// Create an instance of FirebaseObjectBasedPositioning
const positioning = new FirebaseObjectBasedPositioning(firestore);

Changing default settings

// Change the default collection name
positioning.setCollectionName('new-collection-name');

// Change the current object detection model
positioning.setCurrentModel('mobilenet v2');

// Add a custom model
const CustomModelComponent = React.lazy(() => import('../components/CustomModelCamera'));
positioning.addCustomModel('custom-model', CustomModelComponent);

// Change the current positioning method
positioning.setCurrentPositioningMethod('custom-positioning-method');

// Set the maximum distance to detect objects
positioning.setMaxDistanceToDetectObjects(20); // 20 meters

// Set the maximum heading to detect objects
positioning.setMaxHeadingToDetectObjects(90); // 90 degrees

Registering new object

// Example of registering an object
const tensorCameraResult = { /* TensorCameraResult data here */ };
const extraData = { /* Any extra data here */ };

positioning.onRegisterObject(tensorCameraResult, extraData)
  .then(() => {
    console.log('Object registered successfully.');
  })
  .catch((error) => {
    console.error('Error registering object:', error);
  });

Unregistering existing object

// Example of unregistering an object
const objectId = '1234567890';

positioning.onUnregisterObject(objectId)
  .then(() => {
    console.log('Object unregistered successfully.');
  })
  .catch((error) => {
    console.error('Error unregistering object:', error);
  });

Retriving objects

// Example of getting all registered objects with some condition
const conditions = { /* Your conditions here */ };

positioning.getRegisteredObjects(conditions)
  .then((objects) => {
    console.log('Registered objects:', objects);
  })
  .catch((error) => {
    console.error('Error getting registered objects:', error);
  });

// Example of getting nearby objects
positioning.getNearbyObjects()
  .then((nearbyObjects) => {
    console.log('Nearby objects:', nearbyObjects);
  })
  .catch((error) => {
    console.error('Error getting nearby objects:', error);
  });

Get React Component

!NOTE More information about these Components can be found here.

The documentation presents a collection of React components available for real-time object detection and classification using pre-trained TensorFlow.js models. These components are designed to seamlessly integrate into React Native applications and leverage the device's camera capabilities to detect and classify a variety of objects. Each component provides a straightforward interface for configuring and receiving detection or classification results. They allow the phone's camera to view the environment and send images for processing by the object recognition model, making it easy to integrate into existing projects and customize according to specific application needs.

// Change the current object detection model
positioning.setCurrentModel('mobilenet v2');

// Get Camera Component for Mobilenet v2
const Component = positioning.getReactCameraComponent();

const App = () => {
  const handleNewResults = (results) => {
       console.log('New detection results:', results);
       // Your logic here
     };

  return (
    <Component
      minPrecision={25}
      numberOfFrames={150}
      onNewResults={handleNewResults}
      availableObjects={['person', 'car']}
    />
  );
}