Madhura Phatak

Papers

Total Citations

H-Index

About

Madhura Phatak is a rising researcher in computer vision and embodied AI, whose work focuses on bridging the gap between human-robot coexistence and scene understanding. Her key research areas include synthetic dataset generation, human-object interaction (HOI) detection, and scene graph construction for autonomous agents. Phatak’s most notable contribution is the creation of **HOIverse**, a synthetic scene graph dataset introduced in 2025 that captures detailed human-object interactions. This dataset addresses a critical gap in existing research by providing reliable, scalable annotations for training agents to localize and identify human actions in shared environments—enabling downstream tasks like navigation and planning. While still early in her career, with her first citation already recorded, HOIverse represents a foundational step toward more robust, real-world robotic perception. Phatak’s work is particularly significant for researchers developing socially aware robots that must interpret human behavior in dynamic settings. By prioritizing synthetic data generation, she offers a cost-effective, controllable alternative to labor-intensive real-world annotation, positioning her as an emerging voice in the future of human-robot interaction and scene understanding.