Zhiheng Ma

Papers

Total Citations

H-Index

About

Zhiheng Ma is a researcher whose work lies at the intersection of computer vision and few-shot learning, with a particular focus on advancing object counting and segmentation capabilities. His most notable contribution is the pioneering study "Can SAM Count Anything? An Empirical Study on SAM Counting" (2023), which rigorously investigates the application of Meta AI's Segment Anything Model (SAM) to the challenging task of few-shot object counting. This work systematically evaluates SAM's ability to count objects from unseen categories, revealing both its strengths and limitations in this domain. With 10 citations already, this paper has quickly become a reference point for researchers exploring the intersection of foundation models and dense prediction tasks. Ma's research addresses a critical gap in computer vision: enabling models to count novel objects with minimal supervision. His empirical approach provides valuable insights into the transferability of large-scale segmentation models, offering practical guidance for future developments in few-shot counting and beyond. Through this work, Ma has established himself as a thoughtful contributor to the ongoing dialogue about how generalist vision models can be adapted for specialized, data-efficient tasks.