https://github.com/danny-1k/livecell

https://github.com/danny-1k/livecell

Last synced: 3 months ago
JSON representation

• Host: GitHub
• URL: https://github.com/danny-1k/livecell
• Owner: danny-1k
• Created: 2023-11-21T09:43:16.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2023-12-25T18:10:39.000Z (over 1 year ago)
• Last Synced: 2024-12-29T07:36:28.640Z (5 months ago)
• Language: Python
• Size: 2.52 MB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
## *very minimal effort* Segnet and Unet for Cell segmentation (Livecell dataset)

## Method

The cell images and segmentation masks are splitted into small patches (both for generalisation and resource efficiency).

The Segmentation models are then trained on a patch-level. Inference is done by "stitching" the generated patch by patch segmentation masks.

## Results

Without hyperparameter tuning or efforts to increase performance, the baseline model acheives a train IOU and DICE of 0.44 and 0.59 respectively and test IOU and DICE of 0.52 and 0.61 respectively.

![](pred/A172_Phase_C7_1_00d00h00m_1.png)