Shivam Singh

Computer Vision & Generative Models

I'm a Computer Science PhD student at Arizona State University working on generative computer vision. My research interests lie specifically in the areas of layered image / video decomposition and editing, unified modeling for generation and understanding, and long horizon world modeling.

Open to research collaborations — reach out by email
Email CV Google Scholar GitHub
Shivam Singh

News

Research

RefEdit teaser
RefEdit: A Benchmark and Method for Improving Instruction-based Image Editing Model for Referring Expression

Bimsara Pathiraja, Maitreya Patel, Shivam Singh, Yezhou Yang, Chitta Baral

ICCV 2025

project page · arXiv · code · bibtex

RefEdit, an instruction-based editing model trained on 20,000 synthetic triplets, outperforms baselines trained on millions of samples in complex scene editing and achieves state-of-the-art results on referring-expression and traditional benchmarks.

Chimera teaser
Chimera: Compositional Image Generation using Part-Based Concepting

Shivam Singh, Yiming Chen, Agneet Chatterjee, Amit Raj, James Hays, Yezhou Yang, Chitta Baral

project page · arXiv

Chimera is a personalized image-generation model trained on a semantic part-based dataset, enabling novel object synthesis by combining parts from multiple images via textual instructions — outperforming baselines by 14% in compositional accuracy and 21% in visual quality.

Projects

Seamless 360° panorama teaser
Seamless 360° Panoramas via Circular MultiDiffusion: Modular Tiling for Continuous Wraparound Generation

2026

code

Extended MultiDiffusion (Bar-Tal et al., 2023) and InfiniteDiffusion (Goslin, 2026) to produce truly seamless 360° panoramas by making tile splat/read modular.

Virtual Insomnia Patient — Clinical Simulation Platform

Funded by DoD — in collaboration with UofA

Jan–May 2026

Developed agents to function as virtual standardized insomnia patients for therapist training and clinical education. Finetuned LLMs for context-aware dialogue systems to simulate realistic patient–therapist interactions.

Experience

Teaching