Publications by authors named "Yu-shen Liu"

Surface reconstruction for point clouds is one of the important tasks in 3D computer vision. The latest methods rely on generalizing the priors learned from large scale supervision. However, the learned priors usually do not generalize well to various geometric variations that are unseen during training, especially for extremely sparse point clouds.

View Article and Find Full Text PDF

Magnetic solid-phase extraction (MSPE) technology for tetracycline (TCC) was developed by employing the novel and pre-designed FeO-COOH@hydrogen-bonded organic frameworks (HOFs) adsorbents in complex food samples. The HOF shell was grown onto the FeO-COOH core by in-situ self-assembled method. The excellent MSPE performances with less solvent, less adsorbent and time consumption were derived from the hydrogen bonding, π-π and hydrophobic interactions between HOF shell and TCC.

View Article and Find Full Text PDF

Novel magnetic covalent organic frameworks (COFs) were prepared by one-pot synthetic strategy and employed as an efficient adsorbent for magnetic solid-phase extraction (MSPE) of naphthaleneacetic acid (NAA) in food samples. Depending on the predesigned the hydrogen bonding, π-π and hydrophobic interactions of magnetic COFs, the efficient and selective extraction process for NAA was achieved within 15 min. The magnetic COFs adsorbent combined with HPLC-UV was devoted to develop a novel quantitative method for NAA in complex food.

View Article and Find Full Text PDF

We propose a novel method called SHS-Net for point cloud normal estimation by learning signed hyper surfaces, which can accurately predict normals with global consistent orientation from various point clouds. Almost all existing methods estimate oriented normals through a two-stage pipeline, i.e.

View Article and Find Full Text PDF

Learning signed distance functions (SDFs) from point clouds is an important task in 3D computer vision. However, without ground truth signed distances, point normals or clean point clouds, current methods still struggle from learning SDFs from noisy point clouds. To overcome this challenge, we propose to learn SDFs via a noise to noise mapping, which does not require any clean point cloud or ground truth supervision.

View Article and Find Full Text PDF

Surface reconstruction for point clouds is an important task in 3D computer vision. Most of the latest methods resolve this problem by learning signed distance functions from point clouds, which are limited to reconstructing closed surfaces. Some other methods tried to represent open surfaces using unsigned distance functions (UDF) which are learned from ground truth distances.

View Article and Find Full Text PDF

The traditional 3D object retrieval (3DOR) task is under the close-set setting, which assumes the categories of objects in the retrieval stage are all seen in the training stage. Existing methods under this setting may tend to only lazily discriminate their categories, while not learning a generalized 3D object embedding. Under such circumstances, it is still a challenging and open problem in real-world applications due to the existence of various unseen categories.

View Article and Find Full Text PDF

Background: Phenolic endocrine-disrupting chemicals (EDCs) are widespread and easily ingested through the food chain. They pose a serious threat to human health. Magnetic solid-phase extraction (MSPE) is an effective sample pre-treatment technology to determine traces of phenolic EDCs.

View Article and Find Full Text PDF

Learning radiance fields has shown remarkable results for novel view synthesis. The learning procedure usually costs lots of time, which motivates the latest methods to speed up the learning procedure by learning without neural networks or using more efficient data structures. However, these specially designed approaches do not work for most of radiance fields based methods.

View Article and Find Full Text PDF

Most existing point cloud completion methods suffer from the discrete nature of point clouds and the unstructured prediction of points in local regions, which makes it difficult to reveal fine local geometric details. To resolve this issue, we propose SnowflakeNet with snowflake point deconvolution (SPD) to generate complete point clouds. SPD models the generation of point clouds as the snowflake-like growth of points, where child points are generated progressively by splitting their parent points after each SPD.

View Article and Find Full Text PDF

Covalent organic framework (COF)-decorated magnetic nanoparticles (FeO@DhaTab) with core-shell structure have been synthesized by one-pot method. The prepared FeO@DhaTab was well characterized, and parameters of magnetic solid-phase extraction (MSPE) for parabens were also investigated in detail. Under optimized conditions, the adsorbent dosage was only 3 mg and extraction time was 10 min.

View Article and Find Full Text PDF

The task of point cloud upsampling aims to acquire dense and uniform point sets from sparse and irregular point sets. Although significant progress has been made with deep learning models, state-of-the-art methods require ground-truth dense point sets as the supervision, which makes them limited to be trained under synthetic paired training data and not suitable to be under real-scanned sparse data. However, it is expensive and tedious to obtain large numbers of paired sparse-dense point sets as supervision from real-scanned sparse data.

View Article and Find Full Text PDF

Hydrogels are a class of biomaterials used in the field of tissue engineering and drug delivery. Many tissue engineering applications depend on the material properties of hydrogel scaffolds, such as mechanical stiffness, pore size, and interconnectivity. In this work, we describe the synthesis of peptide/polymer hybrid double-network (DN) hydrogels composed of supramolecular and covalent polymers.

View Article and Find Full Text PDF

Efficient magnetic solid phase extraction using crystalline porous polymers can find important applications in food safety. Herein, the core-shell FeO@COFs nanospheres were synthesized by one-pot method and characterized in detail. The porous COF shell with large surface area had fast and selective adsorption for propylparaben via π-π, hydrogen bonding and hydrophobic interactions.

View Article and Find Full Text PDF

Point cloud completion concerns to predict missing part for incomplete 3D shapes. A common strategy is to generate complete shape according to incomplete input. However, unordered nature of point clouds will degrade generation of high-quality 3D shapes, as detailed topology and structure of unordered points are hard to be captured during the generative process using an extracted latent code.

View Article and Find Full Text PDF

A nanostructure of InGaN/GaN multiple quantum well (MQW) nanorods (NRs) was fabricated using top-down etching with self-organized nickel (Ni) nanoparticles as masks on the wafer. The optical properties of InGaN/GaN MQW NRs were discussed by experiment and theory from a light absorption perspective. Three-dimensional (3D) optical images of NRs were successfully obtained by confocal laser scanning microscopy (CLSM) for physical observation of the optical phenomenon of InGaN/GaN MQW NRs.

View Article and Find Full Text PDF

Fine-grained 3D shape classification is important for shape understanding and analysis, which poses a challenging research problem. However, the studies on the fine-grained 3D shape classification have rarely been explored, due to the lack of fine-grained 3D shape benchmarks. To address this issue, we first introduce a new 3D shape dataset (named FG3D dataset) with fine-grained class labels, which consists of three categories including airplane, car and chair.

View Article and Find Full Text PDF

Learning discriminative shape representation directly on point clouds is still challenging in 3D shape analysis and understanding. Recent studies usually involve three steps: first splitting a point cloud into some local regions, then extracting the corresponding feature of each local region, and finally aggregating all individual local region features into a global feature as shape representation using simple max-pooling. However, such pooling-based feature aggregation methods do not adequately take the spatial relationships (e.

View Article and Find Full Text PDF

3D shape reconstruction from multiple hand-drawn sketches is an intriguing way to 3D shape modeling. Currently, state-of-the-art methods employ neural networks to learn a mapping from multiple sketches from arbitrary view angles to a 3D voxel grid. Because of the cubic complexity of 3D voxel grids, however, neural networks are hard to train and limited to low resolution reconstructions, which leads to a lack of geometric detail and low accuracy.

View Article and Find Full Text PDF

Industry foundation classes (IFC) is an open and neutral data format specification for building information modeling (BIM) that plays a crucial role in facilitating interoperability. With increases in web-based BIM applications, there is an urgent need for fast loading large IFC models on a web browser. However, the task of fully loading large IFC models typically consumes a large amount of memory of a web browser or even crashes the browser, and this significantly limits further BIM applications.

View Article and Find Full Text PDF

The structural, electrical, and magnetic properties of armchair black phosphorene nanoribbons (APNRs) edge-functionalized by transitional metal (TM) elements V, Cr, and Mn were studied by the density functional theory combined with the non-equilibrium Green's function. Spin-polarized edge states introduce great varieties to the electronic structures of TM-APNRs. For APNRs with Mn-stitched edge, their band structures exhibit half-semiconductor electrical properties in the ferromagnetic state.

View Article and Find Full Text PDF

Learning 3D global features by aggregating multiple views is important. Pooling is widely used to aggregate views in deep learning models. However, pooling disregards a lot of content information within views and the spatial relationship among the views, which limits the discriminability of learned features.

View Article and Find Full Text PDF

Learning 3D global features by aggregating multiple views has been introduced as a successful strategy for 3D shape analysis. In recent deep learning models with end-to-end training, pooling is a widely adopted procedure for view aggregation. However, pooling merely retains the max or mean value over all views, which disregards the content information of almost all views and also the spatial information among the views.

View Article and Find Full Text PDF

We have studied the electronic structure and the current-voltage (I-V) characteristics of one-dimensional InSe nanoribbons using the density functional theory combined with the nonequilibrium Green's function method. Nanoribbons having bare or H-passivated edges of types zigzag (Z), Klein (K), and armchair (A) are taken into account. Edge states are found to play an important role in determining their electronic properties.

View Article and Find Full Text PDF

Purpose: This study develops a novel hybrid (NH) reconstruction plate that can provide load-bearing strength, secure the bone transplant at the prosthesis favored position, and also maintain the facial contour in a mandibular segmental defect. A new patient-match bending technique which uses a three-dimensional printing (3DP) stamping process is developed to increase the interfacial fit between the reconstruction plate and mandibular bone.

Materials And Methods: The NH reconstruction plate was designed to produce a continuous profile with non-uniform thickness and triangular cross-screw patterns with a locking-screw feature at the plate base.

View Article and Find Full Text PDF