Functional Coatings - Object Recognition
Color-based identification for visual AI applications

BASF collaborated with SRI International to develop a vision system that captures color and surface chemistry data to identify consumer packaged goods and apparel articles at the Stock Keeping Unit (SKU) level. The color-based object detection system can also automatically capture and label images to make artificial intelligence models even smarter and more accurate. The system can even identify objects that are partially occluded independent of pose.

BASF's object recognition system consists of digital cameras, LED lighting, software and colors designed to serve as product identifiers. Computer vision end users can utilize the system for improving the object detection performance of visual artificial intelligence (AI) systems for various use cases involving manufactured consumer items as well as labeling their neural network model training image sets.

As a global leader in the chemicals industry, BASF seeks to use its color and surface solutions competencies along with global manufacturing expertise to enable visual AI applications, including e-commerce, smart home and devices, product authentication, recycling and sorting. 

Key properties and features

  • Improve object detection
  • Automated image annotation 
  • Overcome partial occlusions & pose dependency
  • Access color data for visual AI
  • Increased recyclability
  • Digital brand recognition
poc 5 system picture close up white background.jpg
BASF Coatings has the competencies to create value for emerging visual artificial intelligence applications by connecting surface chemistry to color data digital cameras can collect.

Emre Kurtoglu

Functional Coatings - Object Recognition

Object Detection and Dataset Labeling

2021 Embedded Vision Summit

Ian Childers, Head of Technology for Functional Coatings—Object Recognition, discusses object detection and dataset labeling using colors of manufactured objects (preview)