Building Vision Models at Urban Outfitters

This chapter investigates the development and evaluation of machine learning models for image recognition at Urban Outfitters, detailing the transition from in-house solutions to cloud-based AutoML services. The speakers share insights on the challenges of managing large datasets and the importance of model usability and performance in e-commerce. They also discuss creating a unique Dress Data Set for comparison among top computer vision services, while highlighting key methodologies like transfer learning.