POWER > Trends > What's New

IBM PowerAI Vision Lowers Technical Barriers to Create AI Solutions

PowerAI Vision
Whether your enterprise is manufacturing products with high quality, identifying defective insulators on power lines, monitoring railway tracks for maintenance, or analyzing videos for intruders, odds are you acknowledge that there’s a dire need to transform your business with computer vision. You may, however, encounter challenges that hamper your plans to differentiate.
First off, it’s the business users in the enterprise who are the subject matter experts on the data and are well versed with insights that can transform their businesses. Current tools to pre-process and label data for training can be laborious, expensive and error prone. Also, business users are not be trained on technologies like machine learning or deep learning that can accelerate the detection of actionable-insights.

Second, advanced technologies, like deep learning, where models for computer vision are trained by labeled data-sets depend on skilled engineers called “data scientists.” Currently, there’s a huge gap in the supply and demand for skilled data scientists. And, let’s face it–not all data scientists are created equal.
And finally, training tasks are extremely compute intensive and typically need to bank on hardware optimized for deep learning jobs. Even if your enterprise has an approved budget to embark on the transition, you may be confused about which combination of software, hardware, and partner consultants is right for you. This is where IBM PowerAI Vision comes in.

PowerAI Vision is an IBM offering that's designed with business users in mind. It provides a clickable interface where business users bring in their data, label data sets, and train deep learning models that can classify images and detect objects. IBM PowerAI Vision is built on principles of auto-ML, where base models with optimized hyper parameters are trained using “transfer-learning.” The offering engages partially trained models to auto label objects on videos to generate volumes and variety of data sets required for training. Data labeling tasks that took days are now much faster. IBM PowerAI Vision is the only product in the market where Deep Learning is applied for Data Labeling. We call it “DL for DL.”

Workflow – Label, Train and Deploy
Key highlights of PowerAI Vision
  • AI for ALL: Radiologists, supervisors, security inspectors, and others can train models without technical skills in deep learning.
  • Classification and object detection: With the click of a button, models can be trained to categorize images and detect objects in images and videos.
  • Accelerate data labeling: Business users can train models iteratively and automatically label videos to generate data sets for training. Auto labeling saves time and reduces costs around  curation of datasets.
  • Faster insights and more accuracy: AI Vision is optimized for GPUs on IBM Power Systems, backed with IBM Research breakthroughs.
  • Enterprise grade offering: The software-hardware stack is validated for interoperability, with support available from IBM.
Interested? Join me at IBM Systems Technical University, April 29-May 3 in Atlanta to learn more about PowerAI Vision.

Srinivas (Srini) Chitiveli is the Global Product Manager for IBM PowerAI Vision.

comments powered by Disqus



2019 Solutions Edition

A Comprehensive Online Buyer's Guide to Solutions, Services and Education.

Technology’s Gifts

Seven charities that innovate for good

Anonymization’s Murky Waters

Data experts aim to balance privacy risk, research potential

IBM’s Greenest-Yet Data Center

IBM’s LEED gold data center

IBM Systems Magazine Subscribe Box Read Now Link Subscribe Now Link iPad App Google Play Store