Come take advantage of our labeling effort to show off your algorithms for large scale object _detection_ and image categorization! The goal is to better understand performance of algorithms on hierarchical recognition at large scale with an eye toward image search and auto-annotation. In this, the second year, the challenge will focus on image categorization and object detection for 1000 object categories. The large number of categories provides interesting hierarchical structure and is meant to complement PASCAL VOC which concentrates on classification and detection for 20 categories. Training and validation data with image and bounding box labels will be provided from ImageNet as well as from a dataset newly collected expressly for this challenge -- test data will come only from the newly collected set.
In order to make the challenge accessible to the widest possible audience we are releasing a baseline recognition algorithm as well as precomputed descriptors for all images.
Results will be presented at the PASCAL VOC workshop held at ICCV 2011. Prizes may indeed be won. Time lines and more details are available here:
http://www.image-net.org/challenges/LSVRC/2011/
(Data and development kit including evaluation code available at the above website.)
Good Luck!
The ILSVRC2011 team._______________________________________________
Imageworld mailing list
Imageworld@diku.dk
It is possible to choose digest delivery instead
You will receive posts bundled as a digest every 1-2 weeks
You can change your settings at
http://lists.diku.dk/mailman/listinfo/imageworld
Keine Kommentare:
Kommentar veröffentlichen