Caltech-256 Object Category Dataset

TLDR

The original Caltech‑101 dataset was built by selecting categories, downloading images from Google Images, and manually filtering out non‑category images. We introduce a challenging set of 256 object categories containing a total of 30,607 images. Caltech‑256 was assembled like Caltech‑101 but with more than double the categories, at least 80 images per category, no rotation artifacts, and a new clutter category for background rejection, and we benchmark it using two simple metrics and a spatial pyramid matching algorithm. We train an interest detector on the clutter category that rejects uninformative background regions.

Abstract

We introduce a challenging set of 256 object categories containing a total of 30607 images. The original Caltech-101 [1] was collected by choosing a set of object categories, downloading examples from Google Images and then manually screening out all images that did not fit the category. Caltech-256 is collected in a similar manner with several improvements: a) the number of categories is more than doubled, b) the minimum number of images in any category is increased from 31 to 80, c) artifacts due to image rotation are avoided and d) a new and larger clutter category is introduced for testing background rejection. We suggest several testing paradigms to measure classification performance, then benchmark the dataset using two simple metrics as well as a state-of-the-art spatial pyramid matching [2] algorithm. Finally we use the clutter category to train an interest detector which rejects uninformative background regions.