Projects

Neural net drone ID system.

Reads behaviour from a live camera and scores risk in real time, entirely on the device.

Complete Machine learning Hardware
A small quadcopter drone in flight
Accuracy on unseen people 89%
Inference latency <45ms
Live video 30fps
Landmarks per frame 500+

Overview

MotionTrack tracks movement, expression, gesture, posture, gaze and proximity from a live camera. A neural network turns those signals into one risk score and four clear tiers, with no cloud processing.

  • Python
  • PyTorch
  • OpenCV
  • MediaPipe
  • scikit-learn
  • NumPy
  • Pandas
  • FER
  • Matplotlib
  • Joblib
  • yt-dlp
What it reads

Six signals, one score.

Read from every frame and fused into one risk score.

How it was built

8 steps
  1. 1

    Plan

    Signals, risk tiers and speed targets.

  2. 2

    Detect

    Face, hands, pose and gaze with OpenCV and MediaPipe.

  3. 3

    Features

    Landmarks turned into 22 behavioural features.

  4. 4

    Calibrate

    A rolling baseline for each person.

  5. 5

    Data

    Labelled, balanced samples at three risk levels.

  6. 6

    Train

    A PyTorch network that scores four risk tiers.

  7. 7

    Integrate

    Async capture with inference in the background.

  8. 8

    Test

    Accuracy, latency and false positives across people and settings.

Every frame

From landmarks to a risk tier.

Under 45 ms
  1. 500+Landmarks
  2. 22Features
  3. 1Risk score
  4. 4Risk tiers

Model and demo

Neural network architecture and training diagnostics visualization
Network and training diagnostics
MotionTrack live behavioral analysis demo interface
MotionTrack live demo, background removed