Towards a Large-Scale Unbiased Machine Learning Benchmark for Cell Instance Segmentation


This is my final project for the course CPSC 437/539 Intro to Database Systems in Fall 2023.


Website not deployed yet.

Final Report

Please find the final report, compiled in format of a paper, here.


Our objective is to develop a comprehensive quantitative benchmark designed to impartially assess deep learning techniques using open cell segmentation datasets. Our goal is to establish a standard similar to “CIFAR” or “ImageNet” in the realms of histology and cellular biology. So far, we have examined seven datasets, with a range of 30 to 7,000 images and encompassing between 7,000 to 1.2 million cells. Two of the largest datasets have been integrated into our benchmark. We have evaluated ten deep learning methods, selecting two for their ease of use in inference processes. We plan to further refine and expand this project and will ultimately launch a website to facilitate widespread access and community involvement.

Datasets we explored

Demo: interactive comparison of different methods


If you find this final report useful in your research or wish to refer to it, please use the following BibTeX entry.

  author =       {Liu, Chen and Liao, Danqi and Wang, Shuangge},
  title =        {Towards a Large-Scale Unbiased Machine Learning Benchmark for Cell Instance Segmentation: Final Report for CPSC 537 Intro to Database Systems},
  howpublished = {\url{}},
  year =         {2023}