Welcome to the MO106 mushroom image data-set!

The purpose of this data-set is to serve people with such mushroom images which are relevant for the testing of classification algorithms. MO106 contains 29,100 mushroom images in 106 classes. The largest class has 581 elements, the smallest 105, while the average size is 275. Images, freely available for download, have varying size between 97x130 (smallest area) and 640x640 (largest area).

Images were selected and downloaded from https://mushroomobserver.org/, the filtering process is described in article: Norbert Kiss and László Czúni: Mushroom Image Classification with CNNs: a Case-Study of Different Learning Strategies, 12th Int'l Symposium on Image and Signal Processing and Analysis (ISPA 2021).

Running md5sum on MO_106.zip should produce: c9e5e37d5affd95c2e7900ff27254f81 .

Using this dataset in publications, please send an email to czuni AT almos.vein.hu and refer the article.

If you have any questions, remarks, please contact one of the above authors.

Download (circa 2.1 GB)