Context

A good data set for notebook developpers to use various augmentation techniques to build a high F1 score model

Content

This is a data set of images of various types of balls. There are 30 different types (classes) . train set includes 3615 files in 30 subdirectories, one for each type of ball. The training set is not balanced having a different number of images for each class. This was done intentionally so notebook developers could test methods to deal with unbalanced data. The data set is however of high quality where the region of interest (ROI), the ball typically comprises about 50% of the pixels.

Acknowledgements

We wouldn't be here without the help of others. If you owe any attributions or thanks, include them here along with any citations of past research.

Inspiration

Your data will be in front of the world's largest data science community. What questions do you want to see answered?