🧠 HumanActivityDetector

HumanActivityClassifier is a SwiftUI-based iOS application for real-time human activity recognition using CoreMotion sensor data and a CoreML model. The app collects motion data, predicts activities with a sliding window mechanism, and visualizes sensor readings using animated line charts in a modern and elegant UI. I first collect the data using HumanActivityRecoderApp HumanActivityRecoderApp then train the model uisng HumanActivityRecognition repo after that I developed the app and deploy the model on IOS

---

✨ Features

• 📡 Real-time accelerometer and gyroscope data collection
• 🧠 Local, offline human activity prediction using a CoreML model
• 📈 Live sensor data visualization as smooth line graphs
• 🪟 Sliding window prediction (64-sample window with 50% overlap)
• 🧱 Clean MVVM architecture with SwiftUI modular components
• 🎨 Beautiful UI with gradient background and card-style result display

---

🧠 Machine Learning Model

The project uses a custom CoreML model trained to recognize human activities using 6-axis data (accelerometer and gyroscope).

• Input:
  • MLMultiArray of shape [1, 64, 6]
    • Each sample includes: acc_x, acc_y, acc_z, gyro_x, gyro_y, gyro_z
• Output:
  • Predicted label (e.g., "Walking", "Running", "Sitting"....)

---

📱 How It Works

1. Tap Start Prediction to begin collecting sensor data.
2. The app records sensor readings every ~20ms.
3. Every time 64 samples are collected, the data is fed into the CoreML model.
4. The prediction result is displayed and updated in real-time.
5. The most recent accelerometer data is plotted as a live curve.
6. Tap Stop Prediction to halt motion tracking and prediction.

---

📦 Tech Stack

• Swift / SwiftUI
• CoreMotion (sensor access)
• CoreML (Deploy DL model on IOS)

---

🚀 Getting Started

📦 Prerequisites

• Xcode 15 or later
• iOS 16+ target
• Swift 5.9+
• Real iPhone device (CoreMotion does not work on Simulator)

🛠️ Build & Run

git clone https://github.com//HumantwwindeActivityClassifier.git
cd HumanActivityClassifier
open HumanActivityClassifier.xcodeproj

1. Select a physical iPhone device.
2. Run the app from Xcode.

📊 Live Data Visualization

The app uses a custom LineGraphView to visualize accelerometer data:

• X (Red), Y (Green), Z (Blue) lines
• Smooth transitions with Canvas rendering
• Auto-scales based on incoming values

🎬 App UI Overview, Easy Start & Stop Recording

Simple and intuitive UI with clear buttons to start and stop recording activities.

Prediction Selection	Prediction Selection	Stop Predicition

---

👨‍💻 Author

🔗 GitHub - @TWWinde

If you find this project helpful, feel free to give it a ⭐️ star!