Text analysis has been a hot topic in the social sciences over the past decade. In comparison, the volume of visual data, such as social media images, street-view photos, and digitized maps, has grown even more rapidly, yet methods for analyzing them remain underdeveloped. This talk draws on my recent work to illustrate how visual data can be incorporated into social science research. I first discuss work that relies on images for observational studies, constructing large-scale image datasets from social media and street-view imagery and illustrating how they can be used to analyze social events and public infrastructure. I then present ongoing work on using images in experimental studies, showing how recent advances in AI can help us design visual stimuli. Taken together, this work highlights both the opportunities and the challenges of using visual data in social science research.