A method for increasing the accuracy of image annotating in crowd-sourcing

Nurmukhametov OR & Baklanov A (2016). A method for increasing the accuracy of image annotating in crowd-sourcing. Modern Problems in Mathematics and its Applications 1662: 206-214.

[img]
Preview
Text (In Russian)
A method for increasing the accuracy of image annotating in crowd-sourcing.pdf - Published Version
Available under License Creative Commons Attribution.

Download (3MB) | Preview

Abstract

Crowdsourcing is a new approach to solve tasks when a group of volunteers replaces experts. Recent results show that crowdsourcing is an efficient tool for annotating large datasets. Geo-Wiki is an example of successful citizen science projects. The goal of Geo-Wiki project is to improve a global land cover map by applying crowdsourcing for image recognition. In our research, we investigate methods for increasing reliability of data collected during The Cropland Capture Game (Geo-Wiki). In this regard, we performed analysis of all main steps of the crowdsourcing campaign: image processing and aggregation of collected votes. During the research, we used methods of Computer Vision and Machine Learning. This allowed us to increase accuracy of the aggregated votes from 76% to 87%. Copyright © by the paper's authors.

Item Type: Article
Uncontrolled Keywords: Crowdsourcing; Data quality; Geo-Wiki; Image classification
Research Programs: Advanced Systems Analysis (ASA)
Depositing User: Luke Kirwan
Date Deposited: 26 Sep 2016 11:42
Last Modified: 26 Sep 2016 11:42
URI: http://pure.iiasa.ac.at/13833

Actions (login required)

View Item View Item

International Institute for Applied Systems Analysis (IIASA)
Schlossplatz 1, A-2361 Laxenburg, Austria
Phone: (+43 2236) 807 0 Fax:(+43 2236) 71 313