Weitao Tang - Academia.edu (original) (raw)

Papers by Weitao Tang

arXiv (Cornell University), May 12, 2019

Real-time object detection framework plays crucial role in autonomous driving. In this paper, we ... more Real-time object detection framework plays crucial role in autonomous driving. In this paper, we introduce the real-time object detection framework called You Only Look Once (YOLOv1) and the related improvements of YOLOv2. We further explore the capability of YOLOv2 by implementing its pre-trained model to do the object detecting tasks in some specific traffic scenes. The four artificially designed traffic scenes include single-car, single-person, frontperson-rearcar and frontcar-rearperson. YOLOv2 model performs well in detecting single-car once scope of camera distance is within 60 feet. For detecting single person, YOLOv2 handle most of situations when camera distance is within 50 feet. For detecting both person and car, YOLOv2 always detect unsuccessfully if person overlaps with car in photo. In frontcar-rearperson scenario, the "success rate of prediction" goes down sharply in 50 feet camera distance which is shorter than the same situation in frontperson-rearcar scenario.

Tensor decompositions, including CANDECOMP/PARAFAC decomposition (CPD), Tucker decomposition (TKD... more Tensor decompositions, including CANDECOMP/PARAFAC decomposition (CPD), Tucker decomposition (TKD), and tensor train decompositions (TTD), are extensions of singular value decomposition (SVD) for matrices. They are frameworks to decompose images or videos data into bases and coefficients. Due to recent developments in artificial intelligence (AI), tensor decomposition techniques are becoming increasingly important due to its compact representation, fast access , and easy reconstruction. However, tensor decompositions are still challenging in both computations and interpretations because CPD lacks orthogonality, TKD lacks sparsity, and TTD lacks both orthogonality and sparsity. To understand these issues, we evaluate their theoretical and practical limitations induced by the lack of orthogonality and sparsity in existing tensor decomposition methods. To overcome these limitations, a tensor decomposition method with both orthogo-nality and sparsity is proposed. Due to the two properti...

2022 IEEE 9th International Conference on Cyber Security and Cloud Computing (CSCloud)/2022 IEEE 8th International Conference on Edge Computing and Scalable Cloud (EdgeCom)

Machine Vision and Applications, 2021

Journal of Machine Vision and Applications, Jul 1, 2021

Lecture Notes in Computer Science, 2021

Tensor decompositions are becoming increasingly important in processing images and videos. Previo... more Tensor decompositions are becoming increasingly important in processing images and videos. Previous methods, such as ANDECOMP/PARAFAC decomposition (CPD), Tucker decomposition (TKD), or tensor train decomposition (TTD), treat individual modes (or coordinates) equally. Their results do not contain a natural and hierarchical connection between a given tensor and its lower-order slices (e.g., a video and its frames). To overcome the practical limitation of existing tensor decomposition methods, we propose an innovative High-Order Orthogonal Decomposition (HOOD) for arbitrary order tensors. HOOD decomposes a given tensor using orthogonal linear combinations of its lower-order slices. Each orthogonal linear combination will be further decomposed. In the end, it decomposes the given tensor into orthogonal rank-one tensors. For object detection and recognition tasks in highresolution videos, HOOD demonstrated great advantages. It is about 100 times faster than CPD with similar accuracy detection and recognition results. It also demonstrated better accuracy than TKD with similar time overhead. HOOD can also be used to improve the explainability because the resulting eigenimages visually reveal the most important common properties of the videos and images, which is a unique feature that CPD, TKD, and TTD do not have.

ArXiv, 2019

object detection framework plays crucial role in autonomous driving. In this paper, we introduce ... more object detection framework plays crucial role in autonomous driving. In this paper, we introduce the real-time object detection framework called You Only Look Once (YOLOv1) and the related improvements of YOLOv2. We further explore the capability of YOLOv2 by implementing its pre-trained model to do the object detecting tasks in some specific traffic scenes. The four artificially designed traffic scenes include single-car, single-person, frontperson-rearcar and frontcar-rearperson.

2020 19th IEEE International Conference on Machine Learning and Applications (ICMLA), 2020

Tensor decomposition techniques are becoming increasingly important in processing videos with lar... more Tensor decomposition techniques are becoming increasingly important in processing videos with large sizes and dimensions. Under the framework of CANDECOMP/PARAFAC decomposition (CPD), this work studies low-rank sparse tensor approximations (LRSTAs) to higher-order tensors. Both theoretical and practical properties are evaluated for LRSTAs to represent large high-resolution videos. The evaluation brings three major contributions of this work. Firstly, the theoretical connection between CPD for high-order tensors and traditional singular value decomposition (SVD) for matrices are established, and the tensor rank for traditional SVD is defined. This provides a theoretical basis to compare tensor-based approach against matrix-based approach under the framework of tensor decompositions. Secondly, the non-orthogonality of CPD and its implications are revealed. The solution set of an LRSTA can only be used as a whole. Thirdly, a computationally efficient algorithm is developed. Its practical properties are also investigated in object detection and recognition in high-resolution videos. The results of the experiments showed that the proposed algorithm can handle large high-resolution videos very efficiently in terms of memory allocation. Results also revealed that commonly used total variations may not be a good evaluation metric for real world applications in computer vision. LRSTAs should be evaluated using the end goal of the applications, such as the accuracy of object detection and recognition.

Madridge Journal of Nanotechnology & Nanoscience, 2017

Transition-metal dichalcogenides (TMDCs) have emerged as a new class of semiconductors that displ... more Transition-metal dichalcogenides (TMDCs) have emerged as a new class of semiconductors that display distinctive properties at the monolayer thickness. Their electronic and optical properties are of particular interest and importance for applications in optoelectronics as light emitters, detectors, and photovoltaic devices. In the first part of this study, the temperature dependence of the energy gap of TMDCs (MoS 2 , MoSe 2 , WS 2 and WSe 2) for monolayers is discussed. The second part focuses on the determination and analysis of the spectral properties of these materials, at monolayer and bulk, in the range of 1.5-3.0 eV by MATLAB simulations. The optical bandgaps of TMDC monolayers have been simulated from their spectral dependence of the absorption coefficient. Case studies of the simulation of the optical properties of these materials on silicon, gold and fused silica substrates are presented.

ELECTRICAL, ELECTRONIC and OPTICAL PROPERTIES OF MoS2 & WS2 by Weitao Tang Two dimensional materi... more ELECTRICAL, ELECTRONIC and OPTICAL PROPERTIES OF MoS2 & WS2 by Weitao Tang Two dimensional materials such as graphene, boron nitride and transition metal dichalcogenide (TMDCs) monolayers have arisen as a new class of materials with unique properties at monolayer thickness. Their electrical, electronic and optical properties are of great importance for a variety of applications in optoelectronics as light emitters, detectors, and photovoltaic devices. This work focuses on MoS2 and WS2, which are two important members of the TMDC class of materials. The properties of monolayer MoS2 and WS2 are investigated as well as the properties of bulk MoS2 and WS2 to provide an understanding of their significant difference. A detailed investigation of the electrical and electronic properties including temperature dependent resistivity, contact resistance, band structure and electronic excitation are discussed in this work. The temperature dependence of the energy gap for monolayer MoS2 and WS2 i...

Semiconductors, 2019

An overview of the physical, structural, electronic, optical, and electrical properties of Transi... more An overview of the physical, structural, electronic, optical, and electrical properties of Transition Metal Dichalcogenides (TMDCs) and their applications are presented in this chapter. In particular, the sulfides and selenides of molybdenum and tungsten, i.e., MoS2, WS2, MoSe2, and WSe2, are considered. The temperature dependence of the energy gap and simulation of the optical properties of these materials on a variety of substrates are emphasized.