data

    BOLD - Birmingham Object Lighting Database

    The Birmingham Object Lighting Database contains high resolution stereoscopic image pair of objects, surfaces, faces and outdoor scenes photographed under well specified lighting conditions. The aim of the database is to provide a test bed for image processing algorithms. Specifically we are interested in algorithms for extracting intrinsic images, de-lighting, light source estimation and shape from shading. To this end we aim to provide images where the lighting conditions, and surface shapes are known (if possible). We also concentrate on images where shadows fall across changes in texture as these are particularly challenging cases. Images are rendered for the sRGB colour space but with a linear luminance characteristic. Click here for descriptions of the lighting conditions and camera set up, ground truth height maps for the surfaces and MatLab code for converting to different colour spaces. BOLD was funded by the EPSRC. © University of Birmingham 2011, all rights reserved. Images are free for academic use only.

    Use the quick searches below to show example images for of all subjects and download their family zip files or use the full search facility to find specific examples. See search tips for help with searching the database.

    Please complete our survey.

     a face

     an outdoor image

     a surface

     an object

    Search Tips Search tips        Search the database Search the database        Search the database Technical information       

    If you use these images please cite Jiang, X., Schofield, A.J., Wyatt, J.L., (2010) Correlation-Based Intrinsic Image Extraction from a Single Image, in K. Daniilidis, P. Maragos, N. Paragios (Eds.) ECCV 2010, Part IV, LNCS 6314, pp. 58–71, Springer-Verlag, Berlin.

    Some of the surface images were used in our ECCV 2010 paper ECCV 2010 paper "Correlation-based Intrinsic Image Extraction from a Single Image" and its supplementary file. These images can be downloaded below. You can also download MatLab code for the algorithm described in the above paper. You can find out how the images were generated in our ECVP 2010 poster.

    Download ECCV Images

    Download Intrinsic image extraction code ECCV 2010
    Download Shadow Detection code BMVC 2011

    © Copyright University of Birmingham 2011 : Database maintained by the Vision Laboratory : For queries e-mail a.j.schofield at bham.ac.uk.