What's new

The following functions, features, and support have been added for IBM Visual Insights Version 1.2.0:

Included with fix pack 1
After installing Version 1.2.0, it is recommended that you install all available fix packs. Fix pack 1 includes the following functional enhancements. In the documentation, functions that are only available with fix pack 1 are marked with these tags.
Models optimized for speed can be run anywhere, but might not be as accurate as those optimized for accuracy. These models use you only look once (YOLO) V3 and will take several hours to train.

A CoreML asset is generated when you train a model using YOLO V3.

Queue for training jobs
When you submit a job for training, if there are not enough resources available for immediate training, the job is added to a queue. For details, see Training a model.
Move and copy between data sets
You can now move and copy videos and images between data sets. See Creating and working with data sets for details.
Mark deployed models as production
The Production tag helps you track which models are ready for production use. You can now mark models as Production on the Deployed models page. Additionally, the Production and Rejected tags are now visible on the Deployed models page. See Production work flow for more information.
Find assets based on metadata
You can now search for images and videos based on their Exchangeable image file format (Exif) metadata. For details, see Search for assets in a data set.

You can view the metadata associated with an image by opening the image and clicking the Show metadata icon ().

View labels added by inferencing
When you deploy a model, if you specify to save inference results to a data set, those results are labeled as Inferred. You can accept or reject the inferred labels. See Refining a model to learn more.
Included with Version 1.2.0
The following functional enhancements are included with Version 1.2.0:
x86 support
Support for Training and Inference product on the x86 platform with Nvidia GPUs.
Product rename
IBM PowerAI Vision has been renamed to IBM Visual Insights. Additionally:
  • The installation location has changed from /opt/powerai-vision to an IBM standard location: /opt/ibm/vision.
  • Runtime tools have removed powerai- from the tool names. For example, vision-start.sh.
  • Runtime Docker container and Kubernetes pod names have removed powerai- from the name.
Updated URL
After upgrading to Version 1.2.0, use the updated URL to access IBM Visual Insights:
IBM Visual Insights stand-alone URL
https://hostname/visual-insights/, where hostname is the system on which you installed IBM Visual Insights.
IBM Visual Insights with IBM Cloud® Private URL
https://proxyhost/visual-insights-RELEASE/, where proxyhost is the host name of your IBM Cloud Private proxy server, and RELEASE is the name you specified in the Release name field when you deployed the Helm chart.
Pre and post inferencing updates
When deploying a model, you can specify that inference results for images are saved to the data set. See Preprocessing and post-processing for details.
User interface improvements
The user interface has an updated look and feel.
  • The toolbar buttons have been replaced by icons with hover text.
  • The Select checkbox now includes a dropdown menu to let you select all images on the current page.
  • Improved support for working with the user interface on an iPad or small browser window.