Developers

New feature to Watson Visual Recognition improves visual classifiers

Share this post:

Recently, IBMer Andy Trice developed a great application demonstrating how to use drones and Watson Visual Recognition to conduct areal surveys. We thought it would be even more impactful if the drone could get smarter using real time feedback from pilots on the ground. This would enable a truly a symbiotic relationship between cognitive computing and humans.

Today, we are excited to introduce a new feature to Watson Visual Recognition that allows users to continue improving visual classifiers. When a user submits an image and Watson Visual Recognition returns a score, users can now continue inputting new images, giving Watson additional examples to learn from. In layman terms, Watson Visual Recognition can now learn over time, continuously getting better at understanding visual information. Here is an example of how it works.

PICONE

PICTWO

Andy agreed that this would be a great improvement and even went ahead an incorporated the functionality into the demonstration! Replicate for your self using his code.

Get started now! Check out the new Retraining feature here

 

More Developers stories

Behind the Code: Meet Bill Higgins

April 29, 2019 | Developers

We're excited to kick off a new series for the Watson blog, focusing on AI and the developers at IBM Watson. AI is one of the hottest topics in tech, and we want to take a deep dive into AI, but more importantly, for you to get to know the people that help bring Watson to life. To kickoff the series, meet Bill Higgins, distinguished engineer, AI for developers, data and AI.  ...read more