Social Image Systems: From Past to Present

This lecture provides an overview of social image systems, starting with a look at image sharing practices in the 20th century. It delves into research on platforms like Flickr, Instagram, and Snapchat, exploring the role of computer vision in enabling these systems. The lecture also covers visual classification using deep learning models like ResNet, GoogleNet, and VGG, with a focus on the ImageNet dataset. Additionally, it discusses the rise of selfies on platforms like Instagram, the use of biometrics as a service, and the impact of faces on engagement metrics. The lecture concludes with insights from large-scale selfie analysis and the Selfiecity project.