Katherine Xu

I am a second-year PhD student at the University of Pennsylvania advised by Jianbo Shi and affiliated with the GRASP Lab. My research is supported by the NDSEG Fellowship.

Prior to Penn, I received my bachelor's and master's in Computer Science (6-3) at MIT, where I conducted research at MIT CSAIL, MIT Quest for Intelligence, and MIT ESI. I've also had the pleasure of interning at Adobe, Meta, and Honda Research Institute.

Email / Scholar / Twitter / Github

Research

I'm interested in enhancing AI systems for visual recognition and synthesis by integrating causality, controllability, and interpretability. Recently, I have been working on diffusion models and multimodal large language models.

	Amodal Completion via Progressive Mixed Context Diffusion Katherine Xu, Lingzhi Zhang, Jianbo Shi IEEE/CVF Computer Vision and Pattern Recognition Conference (CVPR), 2024 Highlight - 2.8% of submissions, 11.9% of accepted papers paper / project page / code Generating hidden object parts using a pretrained diffusion inpainting model by tackling challenging occlusion and object co-occurrence.
	Good Seed Makes a Good Crop: Discovering Secret Seeds in Text-to-Image Diffusion Models Katherine Xu, Lingzhi Zhang, Jianbo Shi IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2025 NeurIPS Attributing Model Behavior at Scale Workshop, 2024 NeurIPS Foundation Model Interventions Workshop, 2024 paper / code
	Detecting Image Attribution for Text-to-Image Diffusion Models in RGB and Beyond Katherine Xu, Lingzhi Zhang, Jianbo Shi IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2025 NeurIPS Attributing Model Behavior at Scale Workshop, 2024 NeurIPS Safe Generative AI Workshop, 2024 paper / code
	Clim-Seg: A Generalizable Segmentation Model for Heat and Flood Risk Mapping Anushka Ray, Katherine Xu, Norhan Bayomi, John E. Fernandez Climate Risk Management, 2024 paper
	Chemistry Insights for Large Pretrained Graph Neural Networks Katherine Xu, Janice Lan NeurIPS AI for Science Workshop, 2022 paper Analyzing neural networks trained on the Open Catalyst dataset by comparing their predictions with chemical intuition. Work done during an internship at Meta.
	A Novel Digital Algorithm for Identifying Liver Steatosis Using Smartphone-Captured Images Katherine Xu, Siavash Raigani, ..., Katherine Fairchild, Leigh Anne Dageforde Transplantation Direct, 2022 paper / code

Teaching and Service

Teaching Assistant, CIS 6800 Advanced Topics in Machine Perception (Fall 2023, Fall 2024)

Co-Organizer, GRASP Lab Student, Faculty, Industry (SFI) Seminar Series (Spring 2024 - present)

Co-Organizer, GRASP Lab Social Committee (Fall 2023 - Spring 2024)

Reviewer, ECCV 2024, WACV 2025