Takes a request, makes a request
You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
Pünkösd Marcell dbb266c744 Added readme 1 year ago
k8s initial commit 2 years ago
svm_prefilter_service Fixed model not downloading 1 year ago
.dockerignore initial commit 2 years ago
.drone.yml upload to docker hub 1 year ago
.gitignore initial commit 2 years ago
Dockerfile use public base image 1 year ago
README.md Added readme 1 year ago
requirements-dev.txt initial commit 2 years ago
requirements.txt Added reporter 2 years ago
uwsgi.ini Finally fixed 2 years ago


SVM Prefilter Service

The purpose of this service is to bring the SVM Ai from the Birbox to the Edge.

This service has the same interface as the input service. Upon receiving a request it enqueues it, runst the Ai prediction, if a bird chirp sound is predicted then this service forwards the request to the original input service in the cloud.

This service needs to communicate with the input service and the model service. And optionally it can report its queue length to birb-latency-collector.


This service can be configured trough envvars:

  • SECRET_KEY: Flask secret key. See https://flask.palletsprojects.com/en/2.0.x/config/#SECRET_KEY
  • SENTRY_DSN: Sentry DSN. See https://docs.sentry.io/product/sentry-basics/dsn-explainer/
  • RELEASE_ID: Release ID for Sentry.
  • RELEASEMODE: Release mode for sentry.
  • MODEL_INFO_URL: URL of the model service endpoint when the information about the used SVM model can be fetched (e.g.: https://birb.k8s.kmlabz.com/model/svm/$default)
  • INPUT_SERVICE_URL: URL of the real input service (e.g.: https://birb.k8s.kmlabz.com/input)
  • REPORT_URL: URL of the birb-latency-collector service to report queue length to (e.g.:
  • REPORT_INTERVAL: Seconds between reporting queue length to birb-latency-collector.