AEye

www.aeye.space

A project for the Disrupt Conference 2016 Hackathon.

AEye helps the visually impaired find objects by using a camera phone as an artifical eye. This is achieved with the use of the IBM Watson Visual Recognition APIs. We also use the Watson Speech to Text service to handle voice input.

Project Layout

/ios

The /ios directory contains the Swift code for the iOS mobile app. See ios/Stevie/Stevie/AppController.swift for the use of Watson's SpeechToText API.

/api

The '/api' directory contains Python web services code used to wrap the Watson Visual Recognition API. You'll also find the code to create the image classifiers and sample image data here.

API

Class list service

Get a list of supported object classes

Example request:

GET /class_list HTTP/1.1

Example response:

{
  "class_list": [
    "keys",
    "oreo",
    "card"
  ]
}

Find service

Find an object class in a given image. Given a specific object class, this endpoint returns a score which represents the likelihood of the image containing that object.

Example request:

POST /find?class=<class> HTTP/1.1
Content-Type: image/jpeg


...JPEG...

Parameter	Required?	Value	Description
class	yes	keys	The thing to look for

Example response:

HTTP/1.1 200 OK
Content-Type: application/json

{
  "class": "keys",
  "score": 0.8
}

Field	Value	Description
class	keys	The class being searched for
score	0.8	Confidence item of that class is present (0.0 - 1.0)

UX Flow