docs

Documentation

Table of Contents

1. System Monitor
2. Smart Car Control
3. Face Detection Lite
4. Webcam Face Detection
5. Online Transfer Learning
6. Webcam Object Detection

System Monitor

This is an app to monitor some key features of my raspberry pi server.

Codes: https://github.com/mohanliu/pi-home-server-webpage/blob/main/html/monitor.php

Codes: https://github.com/mohanliu/pi-home-server-webpage/blob/main/html/monitor_history.php

Features

• Utilized MySQL database to save historical data; set up cron jobs to dump data on every minute
• Using PHP to query data and plot figures using Google Charts
• Live data refreshes every 5 seconds; historical data showing hourly averages

Future Plans

• Monitor other metrics
• Improve front-end design

Smart Car Control

This is a tool kit to control my smart car equiped with a raspberry pi by itself.

Codes: https://github.com/mohanliu/raspberry-pi-car-module

Features

• The smart car hosts a server by itself that can be accessed from the same local network.
• Movements of the car can be controled by a web-based app. This app is optimized for both desktop and mobile devices.
• The backend of the app is a flask server. AJAX is used to execute python functions without a POST submission.

Face Detection Lite

A face detection app with uploaded images

Codes: https://github.com/mohanliu/face_detection_lite

Features

• Customized from official face-api.js.
• Able to switch between heavy and light models.

Webcam Face Detection

A real-time face detection App

Codes: https://github.com/mohanliu/pi-home-server-webpage/tree/main/face_detection_demo

Features

• Optimized for mobile devices which allows both front and back cameras
• Improved front-end design to allow live output feature selection
• Run the app using flask with SSL certificates

Online Transfer Learning

A transfer learning web app for realtime image classification

Codes: https://github.com/mohanliu/pi-home-server-webpage/blob/main/html/transferlearning.php

Features

• Using mobilenet model weights as backbone to extract image features; a KNN model as top layer to transfer to a different target task

Webcam Object Detection

A real-time multi-instance object detection app

Codes: https://github.com/mohanliu/pi-home-server-webpage/blob/main/html/object-detection.php

Features

• Real-time multi-object inference using tensorflow.js
• Apache 2 server with ssl followed instructions from here