Migrating to Visual Recognition from AlchemyVision

On May 19, 2016, the AlchemyVision service was deprecated and its functionality became a part of Visual Recognition (GA). To migrate a Bluemix application from the AlchemyVision service to the Visual Recognition service, note the following changes and update your code accordingly.

  • Classes and classifiers: In AlchemyVision, analyzed images were labelled with "tags". In Visual Recognition, "tags" are called "classes". Groups of classes are called "classifiers". All default tags from AlchemyVision are still available as classes in Visual Recognition.
  • Custom classifiers: Visual Recognition allows you to train and create custom classifiers and classes.
  • Batch input: In Visual Recognition, you can now classify batches of images or image URLs by submitting image files in a compressed (.zip) file or a single image URL in a JSON string.
  • Specify language: In AlchemyVision, you specified the language of a tagging call as a query parameter. In Visual Recognition, you specify the language of a classify call in the Accept-Language header.
  • Endpoints: The following AlchemyVision functionality is included in new endpoints in the GA release of Visual Recognition:
Functionality AlchemyVision endpoint (deprecated) Visual Recognition endpoint (GA)
Tag images with built-in classifiers POST /image/ImageGetRankedImageKeywords
GET /url/URLGetRankedImageKeywords
POST /v3/classify
GET /v3/classify
Detect faces, age range, and gender POST /image/ImageGetRankedImageFaceTags
GET /url/URLGetRankedImageFaceTags
POST /v3/detect_faces
GET /v3/detect_faces
Extract main image from HTML or a website URL POST /html/HTMLGetImage
GET /url/URLGetImage
While you can supply a website URL to the /v3/classify, /v3/detect_faces, and /v3/recognize_text methods to analyze the main image, this functionality is not available as a standalone endpoint in Visual Recognition.