NEC Animal Dataset

Description

NEC Animal dataset consists of a sequence of about 5000 images from 60 toy animals taken at different poses. The pose changes continuously by putting each toy on a turntable. About 72 images are available per object. This dataset can be useful for performance evaluation of pose invariant object recognition or object categorization methods.

Download

Downloading and using this dataset is free for research purposes. The original sequence (316 MB) contains frames where an operator swaps successive animals on the turntable. The clean sequence (281 MB) is the same but with swapping frames eliminated. Recognition results on this dataset is reported in the following paper:

Hossein Mobahi, Ronan Collobert, Jason Weston, "Deep Learning from Temporal Coherence in Video", Proceedings of the 26th Annual International Conference on Machine Learning (ICML'09) , p.737-744, June 14-18, 2009, Montreal, Quebec, Canada, June 2009.