[ Download] [ Browse] [ Intro] [ Details]
[ Benchmarks] [ References] [ Contact]
[ Caltech 101] [ Other Datasets] [ Vision Home Page]
[ Tech Report ]



D o w n l o a d Collection of all 30607 images:   256_ObjectCategories.tar   (1.2 GB)  md5sum

B r o w s e View all the images . Please do not link directly to these images: it overloads our server.

I n t r o A brief introductory online slide show which you can also download as powerpoint slides .

D e t a i l s See the tech report for details.

B e n c h m a r k s Dotted lines show current Caltech-101 benchmarks.
Solid lines show Caltech-256 benchmarks. Please send us yours!

Note: for the purpose of comparing Caltech-101 and Caltech-256 performance (see purple line) we re-implemented the Pyramid Matching algorithm of Lazebnik et. al.   In Caltech-101 we remove "BACKGROUND" and also the "faces-easy" category since it is redundant with "faces".
R e f e r e n c e s Please cite as:
Griffin, G. Holub, AD. Perona, P. The Caltech-256, Caltech Technical Report.
Paper may be downloaded here.
Bibitex entry here.

C o n t a c t Questions, comments, benchmarks and references can be sent to the address below.


Greg Griffin
Last modified: Fri Apr 27 14:47:18 PDT 2007