This coursework concerns the automated classification of images through machine learn- ing techniques. You will work on human kidney tissue cell image data, where training sam- ples and their ground truth are provided. You will develop suitable classification techniques to classify unseen examples, a poster to introduce the problem. You will need to submit a Jupyter notebook and your prediction outcome of a test dataset by 10am on Wednesday 20th July. You will then be provided the ground truth label of the test dataset, available to you from 10am on Wednesday 20th July, which you will use to evaluate the test-time performance. Your poster submission is due at 5pm on Wednesday, 20th July. And the presentation will take place during the evening sessions on Wednesday the 20th.

These are datasets about human kidney cortex cell image, and there are 8 types of cells in this dataset need to be classified. The data were extracted from the Broad Bioimage Benchmark Collection (BBBC) image sets and each image was revised into 28 x 28 gray- scale images.

See here.

[1] A. Woloshuk et al. In situ classification of cell types in human kidney tissue using 3dnuclear staining.Cytometry Part A, 99(7):707–721, 2021

[2] Lisha Li∗, Kevin Jamieson∗∗, Giulia DeSalvo†, et al. Hyperband: Bandit-Based Configuration Evaluation for Hyperparameter Optimization.