I’m excited to announce the availability of a new code pattern, Optimize your visual recognition classification.
The opportunities to build image analysis applications have increased tremendously, including helping to implement self-driving cars, helping the visually impaired to independently navigate large metropolitan cities, and helping insurance companies respond to national disasters more quickly. As one can imagine, with opportunities comes challenges. One such challenge that comes to mind is ensuring that these visual recognition applications are built to accommodate for the growing number of huge image datasets.
With all of these opportunities in mind, I was on a mission! I had not used the visual recognition service very much before, so I was looking to deepen my skillset. I was especially curious about learning how to consume the IBM Cloud services in a stand-alone application. I had heard about the flexibility of utilizing the IBM Cloud client libraries in a non-server deployment model.
As it turns out, the process was relatively easy. I created and trained my first image classifier, and then I had to do couple of manual steps such as capture a set of images and package those images into a zip file. Eureka! What about the notion of automating the training process and using the IoT Platform to speed up the classification step? From there, I came up with the vision to build a JRE-based system application, which can do all the processing automatically. I made the process parameters configurable through a “.properties” file. Users simply need to provide services credentials and the number of classifiers.
As a result of these investigations, I’ve created the pattern, Optimize your visual recognition classification. I hope this pattern helps you get started with the Watson Visual Recognition service and IoT object classification. I’d love to find out your experiences, so leave me a comment below or contact me through GitHub.