Our new code pattern, Enrich multimedia files using services, is now available. Processing multimedia files by using Watson services typically involves using multiple services. For this pattern, we use the following services:
  • Watson Visual Recognition (VR)
  • Watson Speech-to-Text
  • Watson Natural Language Understanding
  • Watson Tone Analyzer
  • Cloudant NoSQL DB
If you’re using a trial IBM Cloud account, you need to be aware that there are some service limitations that you can run into:
  • Watson Tone Analyzer is limited to 2500 API calls.
  • Watson Visual Recognition is limited to 250 API calls in a 24-hour period.
If you run into the limit with the Watson Tone Analyzer service, the solution is a bit of a pain, but also straightforward. Simply delete your current service and create a new one. Of course you will also need to update you environment credentials to match the ones created for the new instance of the service. For the Watson Visual Recognition service, the solution isn’t so simple. If you delete/re-create your service, the API Key you are assigned will be the same key from the previous version of the service. You can’t get a new key until the 24-hour window has expired, so you might as well just keep the service instance you have and wait until the next day to process more multimedia files. The best way to minimize the chance of hitting the limit is to be careful about the number of files you try to process. The Visual Recognition limit of 250 API calls really means it can process 250 images. In the multimedia nalyzer journey, the default frequency of captures is once per every 10 seconds of video. This means that for a 10-minute video, the Visual Recognition service will process 60 images. Following are some tips to help you limit your Visual Recognition API calls:
  • Use short videos. Note that the sample video “Grid Breakers” found in the journey’s media_files directory is around 8 minutes long.
  • Only use Visual Recognition after you have confirmed the video doesn’t have any issues that would prevent it from being completely processed. To do this, do not use the -V option in the process command. For example:
    # Speech-to-Text only
    bin/processMedia -S -f public/media_files/grid-breakers.mp4
  • Reduce the number of seconds between screen captures. For example:
    # Speech-to-Text and VR, image every 20 seconds
    bin/processMedia -S -V -r 20000 -f public/media_files/grid-breakers.mp4
By following these tips, you’ll be able to run your Visual Recognition API calls with fewer issues. Check out the Enrich multimedia files using services developer journey, and let me know how you make out with your files by leaving me a comment below.

1 comment on"Running a Watson Visual Recognition app on a trial IBM Cloud account"

  1. Venkat Gopi April 12, 2018

    Hi I have been working upon IBM Content Management System. Anyone can help me out . I’m working upon Front end UI part, My question is i need some call back methods while searching anything from Searchengine. it should point out the IBM content Manager

Join The Discussion

Your email address will not be published. Required fields are marked *