Dimitris Metaxas

Rutgers, The State University of New Jersey

Author Statistics

Papers

Citation

H-Index

i-10 index

Research Trends

Author Order

Document Type

Co-Authors

Leon Axel

NYU Langone Health

Shaoting Zhang

Shandong First Medical University

Xi Peng

University of Delaware

Junzhou Huang

The University of Texas at Arlington

L. Zhao

East University Of Heilongjiang

Ligong Han

Rutgers Sexual and Reproductive Health and Rights

Pengxiang Wu

Snap (United States)

Xiaolei Huang

Pennsylvania State University

Zhennan Yan

Princeton University

Qingshan Liu

Tianjin Medical University

Cooperative Institutions

Rutgers, The State University of New Jersey

129

Rutgers Sexual and Reproductive Health and Rights

Chinese Academy of Sciences

University of Pennsylvania

Johns Hopkins University

Shanghai Jiao Tong University

Google (United States)

Fudan University

University of Chinese Academy of Sciences

Tsinghua University

Author Statistics

Papers

Citation

H-Index

i-10 index

Research Field

SVDiff: Compact Parameter Space for Diffusion Fine-Tuning

arXiv (Cornell University) (2023)

Ligong Han Yinxiao Li Han Zhang Peyman Milanfar Dimitris Metaxas

Diffusion models have achieved remarkable success in text-to-image generation, enabling the creation of high-quality images from text prompts or other modalities. However, existing methods for customizing these models are limited by handling multiple personalized subjects and the risk of overfitting. Moreover, their large number of parameters is inefficient for model storage. In this paper, we propose a novel approach to address these limitations in existing text-to-image diffusion models for personalization. Our method involves fine-tuning the singular values of the weight matrices, leading to a compact and efficient parameter space that reduces the risk of overfitting and language drifting. We also propose a Cut-Mix-Unmix data-augmentation technique to enhance the quality of multi-subject image generation and a simple text-based image editing framework. Our proposed SVDiff method has a significantly smaller model size compared to existing methods (approximately 2,200 times fewer parameters compared with vanilla DreamBooth), making it more practical for real-world applications.

Overfitting

Modalities

10.48550/arxiv.2303.11305

Cite

Citations (8)

Computer-Aided Diagnosis of Mammographic Masses Using Scalable Image Retrieval

IEEE Transactions on Biomedical Engineering (2014)

Menglin Jiang Shaoting Zhang Hongsheng Li Dimitris Metaxas

Computer-aided diagnosis of masses in mammograms is important to the prevention of breast cancer. Many approaches tackle this problem through content-based image retrieval techniques. However, most of them fall short of scalability in the retrieval stage, and their diagnostic accuracy is, therefore, restricted. To overcome this drawback, we propose a scalable method for retrieval and diagnosis of mammographic masses. Specifically, for a query mammographic region of interest (ROI), scale-invariant feature transform (SIFT) features are extracted and searched in a vocabulary tree, which stores all the quantized features of previously diagnosed mammographic ROIs. In addition, to fully exert the discriminative power of SIFT features, contextual information in the vocabulary tree is employed to refine the weights of tree nodes. The retrieved ROIs are then used to determine whether the query ROI contains a mass. The presented method has excellent scalability due to the low spatial-temporal cost of vocabulary tree. Extensive experiments are conducted on a large dataset of 11 553 ROIs extracted from the digital database for screening mammography, which demonstrate the accuracy and scalability of our approach.

Discriminative model

Scale-invariant feature transform

Tree (set theory)

10.1109/tbme.2014.2365494

Cite

Citations (110)

Sparse shape registration for occluded facial feature localization

Fei Yang Junzhou Huang Dimitris Metaxas

This paper proposes a sparsity driven shape registration method for occluded facial feature localization. Most current shape registration methods search landmark locations which comply both shape model and local image appearances. However, if the shape is partially occluded, the above goal is inappropriate and often leads to distorted shape results. In this paper, we introduce an error term to rectify the locations of the occluded landmarks. Under the assumption that occlusion takes a small proportion of the shape, we propose a sparse optimization algorithm that iteratively approaches the optimal shape. The experiments in our synthesized face occlusion database prove the advantage of our method.

Landmark

Feature (linguistics)

Image registration

10.1109/fg.2011.5771410

Cite

Citations (17)

Open science – combining open data and open source software: Medical image analysis with the Insight Toolkit

Medical Image Analysis (2005)

Terry S. Yoo Dimitris Metaxas

Scope (computer science)

Open research

10.1016/j.media.2005.04.008

Cite

Citations (41)

3D segmentation of rodent brain structures using Active Volume Model with shape priors

Shaoting Zhang Junzhou Huang Mustafa Gökhan Uzunbaş Tian Shen Foteini Delis

Object boundary extraction is an important task in brain image analysis. Acquiring detailed 3D representations of the brain structures could improve the detection rate of diseases at earlier stages. Deformable model based segmentation methods have been widely used with considerable success. Recently, 3D Active Volume Model (AVM) was proposed, which incorporates both gradient and region information for robustness. However, the segmentation performance of this model depends on the position, size and shape of the initialization, especially for data with complex texture. Furthermore, there is no shape prior information integrated. In this paper, we present an approach combining AVM and Active Shape Model (ASM). Our method uses shape information from training data to constrain the deformation of AVM. Experiments have been made on the segmentation of complex structures of the rodent brain from MR images, and the proposed method performed better than the original AVM.

Initialization

Robustness

Active appearance model

Active contour model

10.1109/isbi.2011.5872439

Cite

Citations (7)

A Framework for Recognizing the Simultaneous Aspects of American Sign Language

Computer Vision and Image Understanding (2001)

Christian Vogler Dimitris Metaxas

American Sign Language

10.1006/cviu.2000.0895

Cite

Citations (305)

Left endocardium tracking via collaborative trackers and shape prior

Yan Zhou Shaoting Zhang Nikolaos V. Tsekos Ioannis Pavlidis Dimitris Metaxas

Real-time MRI emerges as a feasible modality for guiding interventions and surgeries on the beating heart. One challenge is the path planning task for intra-cardiac procedures, which requires tracking the inner contour of the left endocardium on the beating heart. We propose a bottom-up approach to solve this tracking problem. On the bottom level, a coarse contour is generated based on a collaborative network of trackers at every time step. On the top level, PCA based shape prior is used to refine the coarse contours, and eventually obtain a smoothed contour of the left endocardium over time. The approach is validated on 14 MRI sequences of a beating heart, with comparisons to the state-of-the-art approaches. Experimental results show that our approach outperforms other approaches, providing a robust, accurate and fast solution to track the left endocardium of the beating heart.

Endocardium

Tracking (education)

BitTorrent tracker

10.1109/isbi.2012.6235665

Cite

Citations (2)

Enabling AI Innovation via Data and Model Sharing: An Overview of the Nsf Convergence Accelerator Track D

AI Magazine (2022)

Chaitanya Baru Michael Pozmantier İlkay Altıntaş Stephen Baek Jonathan D. Cohen

This article provides a brief overview of 18 projects funded in Track D—Data and Model Sharing to Enable AI Innovation—of the 2020 Cohort of the National Science Foundation's (NSF) Convergence Accelerator (CA) program. The NSF CA is focused on transitioning research to practice for societal impact. The projects described here were funded for one year in phase I of the program, beginning September 2020. Their focus is on delivering tools, technologies, and techniques to assist in sharing data as well as data-driven models to enable AI innovation. A broad range of domain areas is covered by the funded efforts, spanning across healthcare and medicine, to climate change and disaster, and civil/built infrastructure. The projects are addressing sharing of open as well as sensitive/private data. In September 2021, six of the eighteen projects described here were selected for phase II of the program, as noted in this article.

Data Sharing

10.1609/aimag.v43i1.19130

Cite

Citations (1)

A Unified Framework of Subspace and Distance Metric Learning for Face Recognition

Analysis and Modeling of Faces and Gestures (2007)

Qingshan Liu Dimitris Metaxas

Sample (material)

10.1007/978-3-540-75690-3_19

Cite

Citations (6)

Enabling AI innovation via data and model sharing: An overview of the NSF Convergence Accelerator Track D

AI Magazine (2022)

Chaitanya Baru Michael Pozmantier İlkay Altıntaş Stephen Baek Jonathan D. Cohen

Abstract This article provides a brief overview of 18 projects funded in Track D—Data and Model Sharing to Enable AI Innovation—of the 2020 Cohort of the National Science Foundation's (NSF) Convergence Accelerator (CA) program. The NSF CA is focused on transitioning research to practice for societal impact. The projects described here were funded for one year in phase I of the program, beginning September 2020. Their focus is on delivering tools, technologies, and techniques to assist in sharing data as well as data‐driven models to enable AI innovation. A broad range of domain areas is covered by the funded efforts, spanning across healthcare and medicine, to climate change and disaster, and civil/built infrastructure. The projects are addressing sharing of open as well as sensitive/private data. In September 2021, six of the eighteen projects described here were selected for phase II of the program, as noted in this article.

Data Sharing

10.1002/aaai.12042

Cite

Citations (1)