Roadmap

Integrated AI workflows with Argo Workflows so users can run end-to-end AI workflows with a single click.

☁️Adding Kubernetes for large-scale compute To handle large datasets and large models, Kubernetes will allow us to scale our compute resources, either on premise or in the cloud. This requires containerizing our applications using Docker, and then orchestrating them using Kubernetes. The MBARI IS department has started on this development.

Adding more algorithms such as semi and self-supervised learning - State-of-the-art methods such as semi-supervised learning have been shown to be effective in improving the performance of models, especially for rare classes. See the MBARI intern paper semi-supervised learning paper for more information on how this was tested on MBARI AUV datasets.

Adding segmentation algorithms for projects that need object area calculations