Zero-human-effort Segmentation

We innovatively proposed an efficient unsupervised learning framework to robustly segment nuclei without human annotations. We first used an iterative training process to improve segmentation quality without human labels. Then we introduced a background boosting technique to enhance the segmentation accuracy. We achieved high fidelity segmentation especially among crowed objects, and IoU improved by 3% compared to original MRCNN.

Keywords: Noisy label, MRCNN, Instant Segmentation, Nuclei segmentation, Crowded Object

• PhD. research project
• Advisors: Dr.Badri Roysam, Dr.Hien Nguyen
• Abstract: GHC Paper
• Poster: GHC 2020 iposter
• Code: [Python] (Keras), partically released

• X. Rebecca. Li, B. Roysam,. Van Nguyen, H., “Toward Zero Human Efforts: Iterative Training Framework for Noisy Segmentation Label”, Grace Hopper Celebration, General Poster, 2020

• Maric, D., Jahanipour, J., Li, X.R. et al. Whole-brain tissue mapping toolkit using large-scale highly multiplexed immunofluorescence imaging and deep neural networks. Nat Commun 12, 1550 (2021). https://doi.org/10.1038/s41467-021-21735-x