The Fundamentals and Applications of Pretrained Vision Models

The Fundamentals and Applications of Pretrained Vision Models

This IDC Perspective focuses on pretrained models for classification and detection and will not cover semantic segmentation or other uses of computer vision. Pretrained vision models enable the rapid deployment of computer vision solutions for companies of all sizes. The accuracy of these models is sufficient for many applications, though attention must be paid toward bias in custom data sets. Edge and embedded solutions will drive down bandwidth demands of cloud computer vision solutions and will help to speed vision-based decision-making frameworks for autonomous systems. While this is a future technology that is available now, the regulatory environment of computer vision applications is evolving, and companies should consider the privacy and other implications of their deployments."Computer vision made great strides in research in the past decade and is poised this decade to make breakthroughs into industry applications. We will see significant growth in this market throughout the 2020s and beyond. We are at the dawn of a new age where computers will regularly perceive and interpret the world around them," said Peter Haas, research director in Computer Vision AI Software at IDC.

Please Note: Extended description available upon request.