ZooProcess accelerates the processing of images of plankton samples produced by the ZooScan instrument through the combined use of classical image segmentation and measurement methods with panoptic segmentation by neural network models. The use of this Artificial Intelligence approach allows for automatic separation of touching objects to enable their exploitation in further steps. This new pipeline supersedes current versions of the software and efficiently handles metadata recording, image acquisition from the scanner, image processing, segmentation, feature extraction and data formatting. The resulting data is imported and managed into EcoTaxa, a web application coupling a database with AI tools to accelerate the labelling of large quantities of plankton images by human operators, who are trained biologists. The operational environment provided by the iMagine AI platform, as well as dedicated training material and support, is available for uptake. The output of this pipeline will contribute to a better understanding of the dynamics of food availability for commercially exploited species and of the effects of climate change.
Addressing target audiences and expressing needs
- Other blended financing
- Use of research Infrastructure
From the RIs providing computing power for iMagine, we would need GPU computing made available after the project ends, if possible, with a commitment over the long term, to be able to build on it. From EMBRC, we will need support to disseminate the product and, in particular, train existing and new users on the new version of the ZooProcess application. We are looking for funding for further development from research/infrastructure projects and from companies, in particular, the one commercialising the ZooScan instrument.
- Public or private funding institutions
- Academia/ Universities
- Private Investors
R&D, Technology and Innovation aspects
Next Steps: Make the wrapper ZooProcess application available. Extend the new ZooProcess application to,
- Create a project and fill its metadata,
- Running the scanner directly from the front end of the application, and
- Uploading the data to EcoTaxa automatically.
In addition, add unit tests to all possible aspects of the application to avoid downstream regressions and enable further development.
The developed pipeline will be installable on the iMagine platform or on other hosted solutions or on personal computers, hence enabling deployment at various scales.
All components are open source (data, code and frameworks) and the results are traceable.
The developed pipeline will be installable on the iMagine platform or on other hosted solutions or on personal computers, hence enabling deployment at various scales.
