/safetypro

Recognize stop signs and other safety-related objects from photos using Microsoft Azure Cognitive Services!

Primary LanguageJavaScript

SafetyPro

Recognize stop signs and other safety-related objects from photos using Microsoft Azure Cognitive Services! Upload an image and find out whether there are safety concerns to be aware of.

  • image detection
  • computer vision

Sample images are provided in the public/img folder.

I kept the layout intentionally simple because it is a tech demo.

Tech Used

  • Machine learning API: Microsoft Azure Cognitive Services
    • Computer Vision
  • JavaScript & EJS
  • HTML / CSS

Take-aways / Optimizations

  • I learned that Azure's off-the-shelf computer vision counts every road sign as "stop sign".
  • This might change or improve over time, and since this app gets classification by API, it will be automatically reflected as changes are made.

Other possible updates may include:

  • Use Cognitive Services for OCR, optical character recognition, to read signage
  • Recall images from a library
  • Open text input to check for custom keywords
  • Overlay rectangles over objects, using CSS
  • Layout improvements

Install

npm install

Things to add

  • Create a config folder with a .env file and add the following as key = value
    • PORT = 2121 (can be any port example: 3000)
    • CLOUD_NAME = your cloudinary cloud name
    • CLOUD_API_KEY = your cloudinary api key
    • CLOUD_API_SECRET = your cloudinary api secret
    • MS_COMPUTER_VISION_SUBSCRIPTION_KEY = your Microsoft Subscription Key
    • MS_COMPUTER_VISION_ENDPOINT = your Microsoft Computer Vision Endpoint
    • MS_FACE_ENDPOINT = your Microsoft Face Endpoint
    • MS_FACE_SUB_KEY = your Microsoft Face Key

Run

  • npm start
  • Go to http://localhost:2122
  • If you need example images to try it out, navigate to public/img/