Roadmap to process new roots data images

[artzet-s]

Assignee to @brhell :

1. Push new image data on github repo via pull request. -> 3 images of différents root size => small, medium, big

2. Write a short result report on the application of existing algorithms with new data : -> push the short report on the docs (images, text) -> push also the code in example sub-directory -> explicit parameters used

3. Describe the problem into sub-problem (like background seperation, denoising, ...) -> write a text and push in the docs

4. Test and compose via existing computer-vision methods: -> test via opencv, scikit-image, scikit-learng, tensor-flow, ... application of their function to the sub-problem -> create issue for each new testing function and push code and results

5. Create news issue with new proposition algorithm and idea to resolve the sub-problem : -> write issue via github -> push code and result proposition

[pradal]

+1

[artzet-s]

I create a wiki page on mini-developer guideline to follow : https://github.com/openalea-incubator/rhizoscan/wiki/Developer-Guideline

[pradal]

Excellent. The best solution will b to create this page on openalea web site to centralise all the guidelines.

[artzet-s]

• Use notebook for demo and script for normal utilisation
• Still try to modify the parameters
• Don't re-scale raw HD image kept intact, and add resize function in the code
• Experiment segmentation methods (Kernel filter, Sobel, etc...)
• Add more images (+2)
• Add reference segmentation image result for testing the last part of the pipeline analysis

[pradal]

Make a new branch names: arabido2 Add a module in pipeline named arabido2.py

Define all the steps you have to implement:

• petri box detection
• image segmentation
• leaf detection