Welcome to the Flip-ML microservice!
Objective

The objective of this microservice is to recognize the orientation of images. This service can recognize orientations of 0° and 180° in scanned documents. In this version, only printed documents can yield reliably accurate results. The output of the service is the predicted class and the probability for each class.

Description

This microservice employs a Deep Learning model of the Convolutional Neural Networks category trained on a modified version of the RVL-CDIP dataset.

https://www.cs.cmu.edu/~aharley/rvl-cdip/, A. W. Harley, A. Ufkes, K. G. Derpanis, "Evaluation of Deep Convolutional Nets for Document Image Classification and Retrieval," in ICDAR, 2015

Features
Classes

The model can recognize the following 2 orientations in a document:

180°
Detections

You can detect the orientation of an image. This image is uploadable through three different ways. Simply click on the buttons below to try.

Specification
Curl command

curl -X GET -L "http://icoservices.kube.isc.heia-fr.ch/flip-ml/specification"

Direct link

http://icoservices.kube.isc.heia-fr.ch/flip-ml/specification