Obtaining suitable training data remains a core challenge in the development of custom computer vision solutions for industrial applications such as those in supply chain and logistics. Industrial data differs fundamentally from common datasets—it is often domain-specific, making it expensive and time consuming to obtain. At the same time, deployed systems generate massive amounts of data. This creates a challenge and an opportunity: How can we efficiently make use of this torrent of data to select and label appropriate images for model training? Vision-language models offer promising potential to address this challenge through their ability to understand and reason about visual concepts with minimal examples. In this talk, we present a generative AI-augmented data curation framework that automates much of the data curation process, minimizing the need for humans in the loop. This approach dramatically reduces annotation effort, accelerates time-to-deployment and reduces the investment needed for developing custom computer vision solutions.


