Skip to content

rfaleiro/Muni-image-recognition

Repository files navigation

Muni-image-recognition | Muni 48 & Noe

App: https://muni-web-app-796082523807.us-west1.run.app/ (off line at the moment)
Project Video: https://www.youtube.com/watch?v=qmK1V2FHKFU

The goal of this project is to track real-time data for San Francisco Municipal (Muni) buses using a local webcam and image recognition to forecast arrival intervals.

Project Overview

  • Webapp to forecast public transportation arrival time and expose frequency trends based on image recognition model.

Data Pipeline & Key Features

The project employs the following pipeline to achieve its goal:

  • Video Capture: A local webcam captures live video of a bus stop.
  • Image Recognition: A YOLO (You Only Look Once) model is used to detect when a new Muni bus arrives at the stop.
  • Data Storage: Detection results are parsed and streamed into a SQLite database.
  • Forecasting: The system provides forecasts based on the trained data.
  • Web Interface: Data is made accessible via a Flask application hosted on Google Cloud Platform (GCP).

Languages, Libraries and Hosting

Based on the repository structure and documentation, the following technologies are utilized:

  • Python
  • YOLO (Image recognition model)
  • Flask (Web framework)
  • Docker (Containerization)
  • HTML (Front-end interface)
  • Google Cloud Platform (Hosting and Cloud SQL)

About

San Francisco Muni bus image recognition and interval forecasting.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors