Radhika Marvin

Papers

Total Citations

H-Index

About

Radhika Marvin is a researcher whose work sits at the intersection of computer vision, audio processing, and human-computer interaction, with a primary focus on active speaker detection. Her major contribution is the creation of the AVA-ActiveSpeaker dataset, a large-scale, meticulously annotated audio-visual resource that filled a critical gap in the field. Prior to this work, the absence of such a dataset constrained progress in speaker diarization, video re-targeting, and speech enhancement. Her foundational paper on this dataset has garnered 19 citations, while its supplementary material has earned 15, demonstrating the community’s reliance on her contributions. By providing a benchmark for active speaker detection, Marvin enabled more robust algorithms for applications ranging from meeting analysis to human-robot interaction. Her work is notable for its practical impact, directly advancing the state of the art in multimodal video analysis. For students and researchers, Marvin’s research exemplifies how careful dataset creation can catalyze progress across multiple downstream tasks, making her a key figure in the development of intelligent, context-aware systems.