This article is an orphan, as no other articles link to it . Please introduce links to this page from related articles ; try the Find link tool for suggestions. (October 2023) |
Xiaoming Liu | |
---|---|
Born | |
Nationality | Chinese-American |
Occupation(s) | Computer scientist and academic |
Academic background | |
Education | B.A., Computer Science and Engineering M.S., Computer Science and Engineering]] Ph.D., Electrical and Computer Engineering |
Alma mater | Beijing Information Technology Institute Zhejiang University Carnegie Mellon University |
Academic work | |
Institutions | Michigan State University |
Xiaoming Liu is a Chinese-American computer scientist and an academic. He is a Professor in the Department of Computer Science and Engineering,MSU Foundation Professor as well as Anil K. and Nandita Jain Endowed Professor of Engineering at Michigan State University. [1]
Liu is most known for his works in the fields of computer vision,machine learning,and biometrics,with a particular focus on facial analysis and three-dimensional (3D) vision. [2] Moreover,he is the recipient of the 2018 and 2023 Withrow Distinguished Scholar Award from the Michigan State University College of Engineering.
Liu is a fellow of the International Association for Pattern Recognition (IAPR) [3] and The Institute of Electrical and Electronics Engineers (IEEE). [4] Additionally,he is the Associate Editor of the journal IEEE Transactions on Pattern Analysis and Machine Intelligence. [5]
Liu completed his Bachelor of Arts degree in Computer Science and Engineering from Beijing Information Technology Institute in 1997. In 2000,he obtained a Master of Science degree in Computer Science and Engineering from Zhejiang University under the supervision of Yueting Zhuang. [6] This was followed by a Ph.D. in Electrical and Computer Engineering,supervised by Tsuhan Chen and Vijayakumar Bhagavatula from Carnegie Mellon University in 2004. [1]
Liu began his academic career in 1998 by joining the Intelligent CAD Lab at Zhejiang University as a research assistant and served until 1999. Between 2000 and 2004,he worked at the Advanced Multimedia Processing Lab at Carnegie Mellon University as a research assistant. In 2012,Liu joined Michigan State University,where he assumed various academic roles over the years. He began as an assistant professor in the Department of Computer Science and Engineering,a position he held from 2012 to 2018,and subsequently became an associate professor in the same department from 2018 to 2020. Since 2020,he has been serving as a professor in the Department of Computer Science and Engineering at Michigan State University. Additionally,he holds appointments as an MSU Foundation Professor since 2021 and as the Anil K. and Nandita Jain Endowed Professor of Engineering at Michigan State University since 2022. [1]
From 2004 to 2012,Liu worked as a Computer Scientist at the Visualization &Computer Vision Lab at GE Global Research. Since 2021,he has held the role of Visiting Researcher Scientist as part of Google's Visiting Researcher Program. He has also been serving as an Associate Editor for the Journal IEEE Transactions on Pattern Analysis and Machine Intelligence since 2023. [5]
With an h-index of 71 and a citation count of 23,105, [2] he has authored over 200 publications spanning the field of computer vision,with a primary focus on recognition,modeling,and 3D perception. [2]
The recognition and analysis of human facial images have been a central focus in the field of computer vision since its inception. Liu's recognition research has focused on fundamental problems such as designing effective loss functions for face matcher learning,integrating identity information across frames,exploring the link between image restoration and recognition in low-quality imagery,and investigating the role of 2D and 3D shapes in recognition. [2] His group designed a margin-based loss function that is adaptive to the diverse image quality in face recognition training dataset,which resulted in a generic face matcher called Adaface. Concentrating his research efforts on tackling video-based face recognition,his study underscored the contribution of temporal dynamics in recognition [7] and introduced an adaptable method called "CAFace" for understanding the connections between identity details across video frames and executing sequential recognition while streaming video. [8]
One of the primary research objectives of Liu's recognition research is to cultivate trust between AI systems and their users. He led research in face presentation attack detection (PAD) and published papers. [9] [10] [11] [12] He created algorithms to mitigate bias in facial recognition by revealing that different demographic groups require distinct convolutional neural network (CNN) kernels,leading to an adaptive architecture that enhances accuracy while reducing bias. [13] [14] Moreover,he proposed a technique Model Parsing for reverse engineering GMs to understand their hyperparameters. [15] His group also developed passive and proactive approaches to deep fake detection and localization. [16] [17]
Liu's modeling research has centered on image alignment and intrinsic image decomposition. He developed Boosted Appearance Model,a discriminative model for image alignment that uses a boosting-based classifier to distinguish between correctly aligned images with ground-truth landmarks and incorrectly aligned images with perturbed landmarks. [18] [19] Additionally,to address the limitations of BAM,he introduced BRM,which focused on learning a score function that is concave in the vicinity of the correct alignment,potentially improving alignment accuracy. Furthermore,he identified the challenge of aligning profile-view faces accurately and addressed this by developing approaches that treat image alignment as a 3D Morphable Models (3DMM) fitting problem,enabling the estimation of 3D facial landmarks. [20] [21] [22]
Liu's intrinsic image decomposition research addressed the problem of 3D reconstruction from 2D images,aiming to estimate high-fidelity 3D surface information of objects or scenes. He proposed an approach based on intrinsic image decomposition,breaking down an image into four components:camera projection matrix,shape parameters,albedo parameters,and illumination parameters,achieved through an encoder. Notably,the absence of ground truth labels for training the encoder leads them to employ two decoders,one for generating 3D shape and another for albedo maps. He used a differentiable renderer to jointly learn the encoder and decoders,ensuring that the rendered image closely matches the original input. [23] [24] Initially developed for faces,this framework serves as a deformable appearance and shape model,akin to the conventional 3D Morphable Model (3DMM) but capable of learning directly from 2D in-the-wild images without the need for 3D scans. Moreover,it was extended to generic objects through the development of implicit functions. [25]
Liu's 3D vision research has explored the fields of 3D object detection and multi-sensor fusion. He proposed a method,referred to as M3D-RPN (Monocular 3D Region Proposal Network) which enhanced the effectiveness of monocular 3D object detection and Bird's Eye View tasks when applied to the KITTI urban autonomous driving dataset. [26] This research was further extended in the direction of video-based detection,the creation of depth equivariant networks and non-maximum suppression (NMS) techniques. [27] [28] Moreover,he specialized in the field of multi-sensor fusion,focusing on integrating cameras with LiDAR or Radar sensors to enhance 3D perception capabilities. His work encompassed tasks like depth completion,velocity estimation,and 3D object detection. [29]
Articulated body pose estimation in computer vision is the study of algorithms and systems that recover the pose of an articulated body,which consists of joints and rigid parts using image-based observations. It is one of the longest-lasting problems in computer vision because of the complexity of the models that relate observation with pose,and because of the variety of situations in which it would be useful.
Object recognition –technology in the field of computer vision for finding and identifying objects in an image or video sequence. Humans recognize a multitude of objects in images with little effort,despite the fact that the image of the objects may vary somewhat in different view points,in many different sizes and scales or even when they are translated or rotated. Objects can even be recognized when they are partially obstructed from view. This task is still a challenge for computer vision systems. Many approaches to the task have been implemented over multiple decades.
Object detection is a computer technology related to computer vision and image processing that deals with detecting instances of semantic objects of a certain class in digital images and videos. Well-researched domains of object detection include face detection and pedestrian detection. Object detection has applications in many areas of computer vision,including image retrieval and video surveillance.
Synthetic data is information that's artificially generated rather than produced by real-world events. Typically created using algorithms,synthetic data can be deployed to validate mathematical models and to train machine learning models.
In computer science,landmark detection is the process of finding significant landmarks in an image. This originally referred to finding landmarks for navigational purposes –for instance,in robot vision or creating maps from satellite images. Methods used in navigation have been extended to other fields,notably in facial recognition where it is used to identify key points on a face. It also has important applications in medicine,identifying anatomical landmarks in medical images.
Deep learning is part of a broader family of machine learning methods,which is based on artificial neural networks with representation learning. The adjective "deep" in deep learning refers to the use of multiple layers in the network. Methods used can be either supervised,semi-supervised or unsupervised.
Convolutional neural network (CNN) is a regularized type of feed-forward neural network that learns feature engineering by itself via filters optimization. Vanishing gradients and exploding gradients,seen during backpropagation in earlier neural networks,are prevented by using regularized weights over fewer connections. For example,for each neuron in the fully-connected layer 10,000 weights would be required for processing an image sized 100 ×100 pixels. However,applying cascaded convolution kernels,only 25 neurons are required to process 5x5-sized tiles Higher-layer features are extracted from wider context windows,compared to lower-layer features.
Adversarial machine learning is the study of the attacks on machine learning algorithms,and of the defenses against such attacks. A survey from May 2020 exposes the fact that practitioners report a dire need for better protecting machine learning systems in industrial applications.
DeepDream is a computer vision program created by Google engineer Alexander Mordvintsev that uses a convolutional neural network to find and enhance patterns in images via algorithmic pareidolia,thus creating a dream-like appearance reminiscent of a psychedelic experience in the deliberately overprocessed images.
Semantic parsing is the task of converting a natural language utterance to a logical form:a machine-understandable representation of its meaning. Semantic parsing can thus be understood as extracting the precise meaning of an utterance. Applications of semantic parsing include machine translation,question answering,ontology induction,automated reasoning,and code generation. The phrase was first used in the 1970s by Yorick Wilks as the basis for machine translation programs working with only semantic representations.
Gregory D. Hager is the Mandell Bellmore Professor of Computer Science and founding director of the Johns Hopkins Malone Center for Engineering in Healthcare at Johns Hopkins University.
Michael J. Black is an American-born computer scientist working in Tübingen,Germany. He is a founding director at the Max Planck Institute for Intelligent Systems where he leads the Perceiving Systems Department in research focused on computer vision,machine learning,and computer graphics. He is also an Honorary Professor at the University of Tübingen.
In the domain of physics and probability,the filters,random fields,and maximum entropy (FRAME) model is a Markov random field model of stationary spatial processes,in which the energy function is the sum of translation-invariant potential functions that are one-dimensional non-linear transformations of linear filter responses. The FRAME model was originally developed by Song-Chun Zhu,Ying Nian Wu,and David Mumford for modeling stochastic texture patterns,such as grasses,tree leaves,brick walls,water waves,etc. This model is the maximum entropy distribution that reproduces the observed marginal histograms of responses from a bank of filters,where for each filter tuned to a specific scale and orientation,the marginal histogram is pooled over all the pixels in the image domain. The FRAME model is also proved to be equivalent to the micro-canonical ensemble,which was named the Julesz ensemble. Gibbs sampler is adopted to synthesize texture images by drawing samples from the FRAME model.
An audio deepfake is a type of artificial intelligence used to create convincing speech sentences that sound like specific people saying things they did not say. This technology was initially developed for various applications to improve human life. For example,it can be used to produce audiobooks,and also to help people who have lost their voices to get them back. Commercially,it has opened the door to several opportunities. This technology can also create more personalized digital assistants and natural-sounding text-to-speech as well as speech translation services.
Video super-resolution (VSR) is the process of generating high-resolution video frames from the given low-resolution video frames. Unlike single-image super-resolution (SISR),the main goal is not only to restore more fine details while saving coarse ones,but also to preserve motion consistency.
Self-supervised learning (SSL) is a paradigm in machine learning for processing data of lower quality,rather than improving ultimate outcomes. Self-supervised learning more closely imitates the way humans learn to classify objects.
A Vision Transformer (ViT) is a transformer that is targeted at vision processing tasks such as image recognition.
Small object detection is a particular case of object detection where various techniques are employed to detect small objects in digital images and videos. "Small objects" are objects having a small pixel footprint in the input image. In areas such as aerial imagery,state-of-the-art object detection techniques under performed because of small objects.