Ringd Measurer: An open source tool for manually generating ring width measurements and a step towards AI generated ring width data.

Ring Measurer is a new tool (currently being tested in-house) for the generation of growth ring width data, both by traditional manual means and novel machine learning methods. Ring

Measurer has been developed to optimise workflows facilitating the ease of generating replicate measurement series and correcting of errors (by adjusting position of meaurements, deleting eronious measurements and inserting new measurements in the middle of existing series).

Currently, the machine learning basedd automated ring measurement tool has been been trained solely on growth rings of the long-lived marine bivalve Glycymeris glycymeris. However, the approach can be used on any images containing rings (i.e. trees, clams, corals or fish otoliths). It should be noteed that given the limited training set, at this point, results will likely be poor for other species (we’ve tested on Arctica islandica with limited success).  The code used for generating the model is publically available on Github and so can be used to generate species specific models to improve results for other species in the future.

Ring Measurer has been tested on Mac OS and Windows based systems. On Windows tha app can be launched directly from an executable file (no interaction with code required) or from a Python sctipt. On Mac OS the app can only be launched from the Python script. No coding is required when launching the app from the script, you jjust have to ensure that the appropriate dependencies are installed prior to running the launch script.

Code for training the model: https://github.com/georgedeath/shellai (available on publication of the manuscript).

Source code for Ring Measurer: https://github.com/ringdater/ring_measurer_source

Please report any bugs or issues, or suggestions for improvements to me via email : d.reynolds2@exeter.ac.uk