Documentation

Find your way of using pictorria.


Getting Started


Overview

Pictorria is a platform for sharing image recognition algorithms. You can run various state of the art image recognition tasks on your images using pictorria api. You can also share your algorithm using pictorria and let many other developers and researchers use it.

Pictorria is designed to be simple and easy to use. To use it, you must sign up, get API key and use the appropriate software development kit. In less than 5 lines of codes, you can process your images ;) that is simple, learn more at Documentation

We are now sharing state of the art Object Detection and Pose Estimation as well as basic Face Detection and scene classification algorithms. Our platform allows for augmenting this services seamlessly into any other application. Please read more about each of the services here.


Registration

Signing up for Pictorria is free but there is an hourly limit of 500 on the number of images you can submit. You can sign up here.


RESTful API

Pictorria uses RESTful API architecture. All request should be send to http://www.pictorria.com/api using POST method with a JSON payload. The JSON string must have four fields.

  • API key: which you obtain by singing up. You can see your API key in your profile.
  • Request type: specifies what is the type of your request. Now we only support recognition request type but soon you will be able to submit learning requests as well.
  • URL: URL of the image you want to be processed. Now we only support requests with URL but soon you will be able to submit request to process files on your local machine.
  • HMAC: We used Hash-Based Message Authentication Code to authenticate messages. To generate the proper HMAC digest for your request, you should Hash the requested URL using SHA-1 standard with the secret key provided to you in your profile.

An example of a request JSON would look like this:

{
"type" : "recognition",
"api_key" : "YOUR API KEY HERE",
"URL" : "LINK TO THE IMAGE",
"HMAC" : "HMAC code generated for the message"
}

Software Development Kits

You can access our API server directly by requesting to www.pictorria.com/api and using the request format mentioned in RESTful API section. To facilitate using our services, we are providing Software Development Kits (SDK) for popular programming languages. Currently we are supporting Matlab and Python but we are going to release SDK for many other popular languages soon.


Python SDK

Here you can find how to use our SDK for python version 2.7

Download

Configuration

Image recognition


Matlab SDK

Download

Configuration

Image recognition


Services

Object Recognition

We are using an implementation of Deformable Part-Based models for object detection. Please find more information about this algorithm in their project page.

Also please cite the following if you use the object recognition algorithm in any work.

@misc{voc-release4,
 author = "Felzenszwalb, P. F. and Girshick, R. B. and McAllester, D.",
 title = "Discriminatively Trained Deformable Part Models, Release 4",
 howpublished = "http://people.cs.uchicago.edu/~pff/latent-release4/"
}

Pose Estimation

We are using Articulate Pose estimation using Flexible mixture of parts for our pose estimation system. You can read more about this work in the project page here:

Also please cite the following if you use the pose estimation algorithm in any work:

@inproceedings{DBLP:conf/cvpr/YangR11,
author = {Yi Yang and Deva Ramanan},
title = {Articulated pose estimation with flexible mixtures-of-parts},
booktitle = {CVPR},
year = {2011},
pages = {1385-1392},
ee = {http://dx.doi.org/10.1109/CVPR.2011.5995741},
crossref = {DBLP:conf/cvpr/2011},
bibsource = {DBLP, http://dblp.uni-trier.de}
}