Image Classification Release Notes



  • Clean up tutorial documentation naming and remove deprecated swagger demo app
  • Rework demonstration page to allow image upload and support object-based detection


  • Clean up documentation for install and parameter descriptions
  • Add documentation and functionality for environment variables in push request


  • Update model to use single image as input type
  • Update javascript demo to run with better CORS behavior (github htmlpreview)
  • Additional documentation for environmental variables
  • Simplify operation for active prediction to use created model (no save+load required)


  • Documentation (lesson1) updated with model runner examples. Deprecation notice in using explicit proto- and swagger-based serves.
  • Update the structure of the protobuf input and output to use flattened (row-based) structure instead of columnar data for all i/o channels. This should allow other inspecting applications to more easily understand and reuse implementations for image data.
  • Update the demonstration HTML pages for similar modifications.


  • Update image examples for open-source video.


  • Documentation and package update to use install instructions instead of installing this package directly into a user’s environment.
  • License addition


  • Refactor to remote the demo bin scripts and rewire for direct call of the script as the primary interaction mechanism.
  • Refactor documentation into sections and tutorials.
  • Create this release notes document for better version understanding.


  • Minor refactor to avoid possibly reserved syntax name


  • Refactor for compliant dataframe usage following primary client library examples for repeated columns (e.g. dataframes) instead of custom types that parsed rows individually.
  • Refactor web, api, main model wrapper code for corresponding changes.


  • Migration from previous library structure to new acumos client library
  • Refactor to not need this library as a runtime/installed dependency


  • Added example for evaluation of a multiple image with all results, saving predictions.