Introduction

The Glasgow Stereo Image Database of Garments is a collection of 80 stereo-pair colour images of garments with corresponding horizontal and vertical disparity maps and mask annotations. This database is based on 16 different off-the-shelve garments. Each garment has been imaged in five different pose configurations using the CloPeMa active binocular robot head. The aim of this database is to serve as a benchmark tool for algorithms for recognition, segmentation and various range image properties of non-rigid objects. This stereo image database is part of a collection of databases published under the EU-FP7 Clothes, Perception and Manipulation (CloPeMa) project (www.clopema.eu).

Technical Description
Images were captured from the CloPeMa robot head. Imaging technology consists of two Nikon DSLR cameras (D5100) at a resolution of 16 mega pixels. See the technical report for a detailed description of the acquisition of the database. Similarly, garments were imaged in five possible pose configurations:
  • flat on the table,
  • folded in half,
  • completely folded,
  • randomly wrinkled; and,
  • hanging over the robot's arm.
Each of the above five configurations were captured synchronously (software-based). Disparities maps were computed using the University of Glasgow's stereo matcher algorithm [1]. Camera calibration parameters (OpenCV compatible) are also included for each captured instance for stereo reconstruction purposes. Matlab 3D reconstruction software is also freely distributed with this database.

Note: Database is divided in 3 7zip files; join them using 7zip.
Click on the slideshow to browse the database

35-mm

Click on the slideshow to browse the database

35-mm without mask