Published Sep 25, 2020

Understanding Cultural Trends with Computer Vision w/ Kavita Bala - #410

Kavita Bala, Dean of Computing and Information Science at Cornell, explores how computer vision can decode global cultural trends and human behavior by analyzing social media data, while addressing privacy concerns and technological advancements in AI for commerce and augmented reality.
Episode Highlights
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence) logo

Popular Clips

Episode Highlights

  • Style Clusters

    Kavita Bala explores the fascinating world of fashion style clusters using computer vision and social media data. By analyzing images from platforms like Instagram, she identifies unique fashion trends across different regions and times of the year. For instance, a distinctive headgear style from Lagos, Nigeria, was identified through this method, showcasing the power of data in revealing cultural specifics 1. Bala explains that some fashion trends, like plaid shirts, are universal, while others are tied to specific events or locations 2.

    We found very distinctive things that people wear in different parts of the world.

    ---

    This approach not only highlights global fashion diversity but also provides insights into societal behaviors and preferences.

       

    Cultural Events

    Bala's research extends beyond fashion to recognize cultural events and societal trends through image data. By leveraging AI tools, she aims to identify significant events, such as political rallies or sports tournaments, based on visual patterns 3. This method offers a new way to understand cultural dynamics and social interactions, providing a deeper insight into how societies function. Bala is particularly excited about applying these techniques to satellite imagery to detect large-scale environmental and societal changes 4.

    You can recognize something as a cultural event versus the sport sporting event.

    ---

    This innovative approach has the potential to transform our understanding of both human and environmental phenomena.

Related Episodes