shravya11/birds-dataset

Caltech-UCSD Birds-200-2011 (CUB-200-2011) Dataset

This is the Caltech-UCSD Birds-200-2011 dataset, containing 11,788 images of 200 bird species, with corresponding binary segmentation masks.

Dataset Structure

The dataset is partitioned using the official train/test split:

• images/:
  • train/: 5,994 JPEG images across 200 species folders
  • validation/: 5,794 JPEG images across 200 species folders
• segmentations/:
  • train/: 5,994 PNG segmentation masks across 200 species folders
  • validation/: 5,794 PNG segmentation masks across 200 species folders

Classes

The dataset contains 200 bird species categories.

Reference

Wah C., Branson S., Welinder P., Perona P., Belongie S. “The Caltech-UCSD Birds-200-2011 Dataset.” Computation & Neural Systems Technical Report, CNS-TR-2011-001.