Searcharxiv⌕ Search

arXiv subjects

Ye Zhu

Publications and source records attributed to Ye Zhu.

98 records · Page 6Linked to original sources
arXiv2018

Multiview Based 3D Scene Understanding On Partial Point Sets

Deep learning within the context of point clouds has gained much research interest in recent years mostly due to the promising results that have been achieved on a number of challenging benchmarks, such as 3D shape recognition and scene semantic segmentation. In many realistic settings however, snapshots of the environment are often taken from a single view, which only contains a partial set of the scene due to the field of view restriction of commodity cameras. 3D scene semantic understanding on partial point clouds is considered as a challenging task. In this work, we propose a processing approach for 3D point cloud data based on a multiview representation of the existing 360° point clouds. By fusing the original 360° point clouds and their corresponding 3D multiview representations as input data, a neural network is able to recognize partial point sets while improving the general performance on complete point sets, resulting in an overall increase of 31.9% and 4.3% in segmentation accuracy for partial and complete scene semantic understanding, respectively. This method can also be applied in a wider 3D recognition context such as 3D part segmentation.

cs.CV↗
arXiv2018

Distinguishing Question Subjectivity from Difficulty for Improved Crowdsourcing

The questions in a crowdsourcing task typically exhibit varying degrees of difficulty and subjectivity. Their joint effects give rise to the variation in responses to the same question by different crowd-workers. This variation is low when the question is easy to answer and objective, and high when it is difficult and subjective. Unfortunately, current quality control methods for crowdsourcing consider only the question difficulty to account for the variation. As a result,these methods cannot distinguish workers personal preferences for different correct answers of a partially subjective question from their ability/expertise to avoid objectively wrong answers for that question. To address this issue, we present a probabilistic model which (i) explicitly encodes question difficulty as a model parameter and (ii) implicitly encodes question subjectivity via latent preference factors for crowd-workers. We show that question subjectivity induces grouping of crowd-workers, revealed through clustering of their latent preferences. Moreover, we develop a quantitative measure of the subjectivity of a question. Experiments show that our model(1) improves the performance of both quality control for crowd-sourced answers and next answer prediction for crowd-workers,and (2) can potentially provide coherent rankings of questions in terms of their difficulty and subjectivity, so that task providers can refine their designs of the crowdsourcing tasks, e.g. by removing highly subjective questions or inappropriately difficult questions.

cs.AI↗
arXiv2016

Revisiting copy-move forgery detection by considering realistic image with similar but genuine objects

Many images, of natural or man-made scenes often contain Similar but Genuine Objects (SGO). This poses a challenge to existing Copy-Move Forgery Detection (CMFD) methods which match the key points / blocks, solely based on the pair similarity in the scene. To address such issue, we propose a novel CMFD method using Scaled Harris Feature Descriptors (SHFD) that preform consistently well on forged images with SGO. It involves the following main steps: (i) Pyramid scale space and orientation assignment are used to keep scaling and rotation invariance; (ii) Combined features are applied for precise texture description; (iii) Similar features of two points are matched and RANSAC is used to remove the false matches. The experimental results indicate that the proposed algorithm is effective in detecting SGO and copy-move forgery, which compares favorably to existing methods. Our method exhibits high robustness even when an image is operated by geometric transformation and post-processing

cs.MM↗
arXiv2016

On the Origins of Near-Surface Stresses in Silicon around Cu-filled and CNT-filled Through Silicon Vias

Micro-Raman spectroscopy was employed to study the near-surface stress distributions and origins in Si around through silicon vias (TSVs) at both room temperature and elevated temperatures for Cu-filled and CNT-filled TSV samples. From the observations, we proved that the stresses near TSVs are mainly from two sources: 1) pre-existing stress before via filling, and 2) coefficients of thermal expansion (CTE) mismatch-induced stress. CTE-mismatch-induced stress is shown to dominate the compressive regime of the near-surface stress distribution around Cu-filled TSV structures, while pre-existing stress dominates the full range of the stress distribution in the CNT-filled TSV structures. Once the pre-existing stress is minimized, the total stress around CNT-filled TSVs can be minimized accordingly. Therefore, compared to Cu-filled TSVs, CNT-filled TSVs hold the potential to circumvent the hassle of stress-aware circuit layout and to solve the stress-related reliability issues.

cond-mat.mtrl-sci↗
arXiv2014

Atomically precise interfaces from non-stoichiometric deposition

Complex oxide heterostructures display some of the most chemically abrupt, atomically precise interfaces, which is advantageous when constructing new interface phases with emergent properties by juxtaposing incompatible ground states. One might assume that atomically precise interfaces result from stoichiometric growth, but here we show that the most precise control is obtained for non-stoichiometric growth where differing surface energies can be compensated by surfactant-like effects. For the precise growth of Sr$_{n+1}$Ti$_n$O$_{3n+1}$ Ruddlesden-Popper (RP) phases, stoichiometric deposition leads to the loss of the first RP rock-salt double layer, but growing with a strontium-rich surface layer restores the bulk stoichiometry and ordering of the subsurface RP structure. Our results dramatically expand the materials that can be prepared in epitaxial heterostructures with precise interface control---from just the $n=\infty$ end members (perovskites) to the entire RP family---enabling the exploration of novel quantum phenomena at a richer variety of oxide interfaces.

cond-mat.mtrl-sci↗
arXiv2012

Determining On-Axis Crystal Thickness with Quantitative Position-Averaged Incoherent Bright-Field Signal in an Aberration-corrected STEM

An accurate determination of specimen thickness is essential for quantitative analytical electron microscopy. Here we demonstrate that a position-averaged incoherent bright-field signal recorded on an absolute scale can be used to determine the thickness of on-axis crystals with a precision of \pm1.6 nm. This method measures both the crystalline and the non-crystalline parts (surface amorphous layers) of the sample. However, it avoids the systematic error resulting from surface plasmons contributions to the inelastic mean free path thickness estimated by electron energy loss spectroscopy.

cond-mat.mtrl-sci↗
arXiv2010

Imaging Grains and Grain Boundaries in Single-Layer Graphene: An Atomic Patchwork Quilt

The properties of polycrystalline materials are often dominated by the size of their grains and by the atomic structure of their grain boundaries. These effects should be especially pronounced in 2D materials, where even a line defect can divide and disrupt a crystal. These issues take on practical significance in graphene, a hexagonal two-dimensional crystal of carbon atoms; Single-atom-thick graphene sheets can now be produced by chemical vapor deposition on up to meter scales, making their polycrystallinity almost unavoidable. Theoretically, graphene grain boundaries are predicted to have distinct electronic, magnetic, chemical, and mechanical properties which strongly depend on their atomic arrangement. Yet, because of the five-order-of-magnitude size difference between grains and the atoms at grain boundaries, few experiments have fully explored the graphene grain structure. Here, we use a combination of old and new transmission electron microscope techniques to bridge these length scales. Using atomic-resolution imaging, we determine the location and identity of every atom at a grain boundary and find that different grains stitch together predominantly via pentagon-heptagon pairs. We then use diffraction-filtered imaging to rapidly map the location, orientation, and shape of several hundred grains and boundaries, where only a handful have been previously reported. The resulting images reveal an unexpectedly small and intricate patchwork of grains connected by tilt boundaries. By correlating grain imaging with scanned probe measurements, we show that these grain boundaries dramatically weaken the mechanical strength of graphene membranes, but do not measurably alter their electrical properties. These techniques open a new window for studies on the structure, properties, and control of grains and grain boundaries in graphene and other 2D materials.

cond-mat.mtrl-sci↗