Multi‑View Datasets

Dataset Overview

The collection aggregates several multi‑view benchmark datasets used for clustering or classification.

Dataset List

AwA

• Description: 30,475 images spanning 50 animal categories.
• Number of Views: 6
• Source: https://cvml.ist.ac.at/AwA/

COIL100

• Description: 7,200 images across 100 categories.
• Number of Views: 3 (Isometric projection, Linear Discriminant Analysis, Neighborhood Preserving Embedding).
• Source: http://www.cs.columbia.edu/CAVE/software/softlib/coil-100.php

MNIST2

• Description: 60,000 digit images (0‑9).
• Number of Views: 3 (Isometric, LDA, NPE).
• Source: http://yann.lecun.com/exdb/mnist/

NUSWIDEOBJ

• Description: 30,000 images across 31 categories.
• Number of Views: 5 (Color histogram, Color moments, Color autocorrelogram, Edge orientation histogram, Wavelet texture).
• Source: https://lms.comp.nus.edu.sg/wp-content/uploads/2019/

PIE

• Description: Subset of the CMU Multi‑PIE face database; 32 × 32 images extracted from sessions C07, C09, C29.
• Number of Views: 3
• Source: https://www.cs.cmu.edu/afs/cs/project/PIE/

YouTubeFaces

• Description: 101,499 facial video frames collected from YouTube.
• Number of Views: Not specified.
• Source: https://www.cs.tau.ac.il/~wolf/ytfaces/

Citation

The dataset is referenced in:

X. Hu et al., "Multi‑View Fuzzy Classification With Subspace Clustering and Information Granules," IEEE Transactions on Knowledge and Data Engineering, vol. 35, no. 11, pp. 11642‑11655, 1 Nov. 2023. doi: 10.1109/TKDE.2022.3231929.