Introductory Workshop: Computer Vision for Social Scientists

When: 27 February 2026, 10.00-15.00

Where: University of Southampton, Highfield Campus, Building 34, room 3011

What:

This one-day workshop introduces social scientists to the systematic analysis of images and videos as data. Visual material is central to politics, communication, and culture, yet many researchers are unfamiliar with the computational tools that allow its study at scale. The workshop offers an accessible entry point to computer vision, highlighting both core methods and their applications in the social sciences.

The detail:

The morning session provides a conceptual overview of computer vision. Participants will be introduced to the basics of deep learning and convolutional neural networks, the foundation of most modern methods. We will review approaches such as image classification, object and face detection, and multimodal techniques that combine text and images. Examples of research using computer vision in social science contexts will be discussed, alongside key ethical considerations, best practices, and infrastructural requirements for large-scale visual data analysis.

The afternoon session is hands-on, focusing on image classification in Python with PyTorch on Google Colab. You will build a machine learning pipeline using an open-source dataset, covering data preparation, model design, training, and evaluation with standard metrics. By the end, you will have trained and tested your own classifier and gained a practical foundation for applying computer vision in research.

The facilitators:

Dr Asra Aslam is Lecturer in Data Science at the University of Sheffield, specialising in AI, machine learning, computer vision, and their applications in health sciences. She has led projects on AI-driven healthcare, collaborated with industry and public bodies such as Transport for London, and received multiple national and international awards recognising her contributions to AI, health, and equity, diversity and inclusion.

Dr Andreu Casas is Associate Professor in Political Communication at Royal Holloway, University of London, and director of the London Social Media Observatory. His research focuses on how digital media shapes policy, collective action, polarisation, and legislative dynamics through innovative computational approaches to text and image data.

Who the training is for:

This training is for doctoral students with some previous programming experience with Python who are wishing to:

• Appreciate how images and videos can be systematically used as data in social science research
• Understand the basic principles behind key computer vision techniques, including image classification and object/face detection
• Gain hands-on experience running beginner-friendly Python code to implement simple computer vision tasks
• Critically assess the opportunities and limitations of applying computer vision methods in different areas of social science
• Identify potential applications of computer vision for their own research projects and future studies.

For this workshop you will use Google Colab and the T4 GPUs in there with the basic free plan. There is the option to install Pytorch and CUDA and use the Jupyter Notebook and for Python beginners to use Jupyter Notebooks. If you can access GPUs all the better!

How to join:

Register your interest by completing the online form here by Monday 16 February 2026.