December 15, 2016 | Written by: Kevin Gong
Categorized: Watson | What's New
Share this post:
Since launching the Visual Recognition API six months ago, one of our most accessible and popular features is general tagging which allows users to understand the contents of images through various keyword tags.
Today, we’re introducing an update to our algorithms that significantly enhances the performance of our general tagging feature. Users will not only see over three times more tags returned for each image on average, but also greater specificity in those tags due to a 150% increase in the active vocabulary of the system. On average, images are analyzed and returned with relevant tags in less than one second.
The results on the left are generated from our previous algorithms, while the results on the right are generated from our updated algorithms.
Visual Recognition for Big Data
Social media analytics company iTrend is already using Visual Recognition to power its data discovery platform. By using general tagging to understand content within images gathered from social media, blogs, and live streaming feeds, iTrend can analyze 20 – 50 times more data than its competition and provide up to 80% of its findings as actionable insights. Read more about their use case here.
All of these changes happen on our end, so you should see the general tagging improvements automatically reflected in your results. Combined with the pricing changes we announced last week, it’s now easier than ever to start building with Visual Recognition.
Seeing What Matters to You
Extend and tailor our general tagging service to meet your business needs with our custom classifier feature, allowing you to custom train Watson to recognize your own visual content. Generate your own high quality tag output with our Visual Training support for creating your own custom classifiers. Try it out in our demo or read about how our partner OmniEarth uses it to help the State of California save water.
Need help or have questions?
We’re excited to see what you build with Watson Visual Recognition, and we’re happy to help you along the way. Try the general tagging feature and share any questions or comments you have on our developerWorks forums.