The Spot-the-diff dataset consists of 13,192 image pairs and corresponding human-provided text annotations that describe the differences between the two images.

Citation

@inproceedings{jhamtani2018learning, title={Learning to Describe Differences Between Pairs of Similar Images}, author={Jhamtani, Harsh and Berg-Kirkpatrick, Taylor}, booktitle={Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing (EMNLP)}, year={2018} }