SERVICE DETAILS
Phytoplankton Classifier
Overview
OSCAR service wrapping the AI4OS phytoplankton classifier for synchronous and MinIO-triggered image classification. Its script calls predict_data with an UploadedFile to work around upstream CLI file-argument incompatibility.
Deployment requirements
- Memory
- 4 GiB
- Processors
- 1 vCPU0 GPU
Files & artifacts
- fdl.yml (opens in a new tab)text/yaml
OSCAR FDL
Show content of fdl.yml
- script.sh (opens in a new tab)text/x-shellscript
Prediction script
Show content of script.sh
- README.md (opens in a new tab)text/markdown
Deployment and validation notes
Show content of README.md
- icon.png (opens in a new tab)image/png
Phytoplankton classifier icon — Original stylized diatom inside image-recognition brackets, representing aquatic microscopy and automated classification.
Show content of icon.png
- sample.jpg (opens in a new tab)image/jpeg
Sample phytoplankton image
Show content of sample.jpg
- Generated prediction JSONExpected artifact
Generated prediction JSON
Usage & acceptance tests
Synchronous classification
Invoke the model with the sample image and verify the synchronous response; a separate decoded JSON check verifies its content.
Call the service synchronously.
oscar-cli service run phyto-plankton-classifier --file-input {input} --output {destination}
Asynchronous classification
Upload the sample image; wait for the bucket-triggered job and retrieve its output.
Upload to the service input prefix.
oscar-cli service put-file phyto-plankton-classifier {source}Wait for Kueue admission, model initialization, and classification.
Time required: PT150SRetrieve the newest prediction output.
oscar-cli service get-file phyto-plankton-classifier --download-latest-into {destination}
RO-Crate metadata
- application/json
Show content of ro-crate-metadata.json