|
This section describes main characteristics of the corel5k data set and its attributes: General information
Additional informationThis data set contains 5000 Corel images. There are 374 words in total in the vocabulary (labels) and each image has 4-5 keywords. Images are segmented using Normalized Cuts. Only regions larger than a threshold are used, and there are typically 5-10 regions for each image. Regions are then clustered into 499 blobs using k-means, which are used to describe the final image. In this section you can download some files related to the corel5k data set:
|
|||||||||||||||||||||||||||||||||
|