The IllusionAnimals dataset is a FiftyOne dataset containing 2k samples, created by researchers from the Iran University of Science and Technology in 2024.Illusory VQA: Benchmarking and Enhancing Multimodal Models on Visual IllusionsThe dataset contains 10 animal categories and one no-illusion category, with an image resolution of 512×512 pixels, and is used to evaluate the ability of multimodal models in identifying and explaining animal-based visual illusions.

Citation

@misc{rostamkhani2024illusoryvqabenchmarkingenhancing, title={Illusory VQA: Benchmarking and Enhancing Multimodal Models on Visual Illusions}, author={Mohammadmostafa Rostamkhani and Baktash Ansari and Hoorieh Sabzevari and Farzan Rahmani and Sauleh Eetemadi}, year={2024}, eprint={2412.08169}, archivePrefix={arXiv}, primaryClass={cs.CV}, url={https://arxiv.org/abs/2412.08169}, }