Dual-mode detection of dopamine based on 0D/2D/2D CuInS2/ZnS quantum dot-black phosphorous nanosheet-TiO2 nanosheet nanocomposites.

In this work, we designed new dual-mode "turn-on" electrochemical (EC) and photoelectrochemical (PEC) sensors for the detection of dopamine (DA) based on 0D/2D/2D CuInS2/ZnS quantum dot (QD)-black phosphorous nanosheet (BPNS)-TiO2 nanosheet (TiO2NS) nanocomposites. QDs can not only improve the photocurrent of the developed PEC sensors, but also provide the electrochemical signal in the EC detection. BPNSs as p-type semiconductor with high conductive properties work as electron acceptors and are utilized to improve the sensitivity of the DA PEC and EC sensors. Under irradiation of visible light or the applied voltage, DA is both excited and releases electrons, realizing "turn-on" detection. The PEC sensors have a linear range of 0.1-100 µM with a lower detection limit of 0.028 µM. For the EC detection, BPNSs can accelerate electron transfer which attribute to its excellent conductivity. In the range of 1-200 µM, the working curve of DA detection by the EC sensors was established and the detection limit is 0.88 µM. Comparing the two methods, the PEC sensors have a lower detection limit, and the EC sensors have a wider monitoring range. The dual-mode sensors of EC and PEC pave an effective way for the detection in biological and medical fields.

A tyrosinase-induced fluorescence immunoassay for detection of tau protein using dopamine-functionalized CuInS2/ZnS quantum dots.

Rapid, highly sensitive detection of tau protein and other neurodegenerative biomarkers remains a significant hurdle for diagnostic tests for Alzheimer's disease. In this work, we developed a novel tyrosinase (TYR)-induced tau aptamer-tau-tau antibody (anti-tau) sandwich fluorescence immunoassay to detect tau protein that used dopamine (DA)-functionalized CuInS2/ZnS quantum dots as the fluorophore. CuInS2/ZnS core/shell quantum dots with high luminescence, low toxicity, and excellent biocompatibility were successfully fabricated and decorated with DA through amide conjugation. Meanwhile, TYR was conjugated with anti-tau by a click reaction. When DA-functionalized CuInS2/ZnS quantum dots were added to the sandwich system, TYR catalyzed the transformation of DA to dopamine quinone, which acted as an effective electron acceptor and triggered fluorescence quenching. The fluorescence intensity of the immunoassay based on DA-functionalized CuInS2/ZnS quantum dots shows good performance in terms of linearity with the logarithm of tau protein concentration, with a linear concentration range from 10 pM to 200 nM. This work is the first to use a TYR-induced fluorescence immunoassay for the rapid detection of tau protein, paving a new way for the detection of disease biomarkers. Graphical abstract.

ZnO flower-rod/g-C3N4-gold nanoparticle-based photoelectrochemical aptasensor for detection of carcinoembryonic antigen.

A highly sensitive and selective photoelectrochemical (PEC) aptasensor was constructed for carcinoembryonic antigen (CEA) detection based on ZnO flower-rods (ZnO FRs) modified with g-C3N4-Au nanoparticle (AuNP) nanohybrids. The nanohybrids of g-C3N4-AuNPs can improve the visible light absorbance of ZnO FRs and enhance the PEC property. We designed a sandwichlike structure formed with DNA hybridization of NH2-probe1, CEA aptamer, and CuS-NH2-probe2 to detect CEA. The p-type semiconductor CuS nanocrystals (NCs) at the terminational part of sandwichlike structure work as electron traps to capture photogenerated electrons and consequently lead to a magnified photocurrent change. The results indicate that the photocurrent is increased when CEA antigen (Ag) is introduced. Since the sandwichlike structure is destroyed, CuS NCs are restricted to capture photogenerated electron. The PEC aptasensor for CEA determination is ranged from 0.01 ng·mL-1 to 2.5 ng·mL-1 with a detection of 1.9 pg·mL-1. The proposed aptasensor exhibits satisfactory PEC performances with rapid detection, great sensitivity and specificity. Specially, this PEC aptasensor shows a reliable result for the determination of CEA in invalid human serum compared with the ELISA method. The designed aptasensor may provide a new idea for a versatile PEC platform to determine various molecules. Graphical abstract ᅟ.

Development of CdSe⁻ZnO Flower-Rod Core-Shell Structure Based Photoelectrochemical Biosensor for Detection of Norovirous RNA.

In this work, the CdSe⁻ZnO flower-rod core-shell structure (CSZFRs) was prepared by ion-exchange method. The surface of CSZFRs was modified by 3-mercaptopropionic acid (MPA), and then the DNA probe was immobilized on the surface via chemical bond between -NH2 of DNA probe and -COOH of MPA. Finally, the target norovirous (NV) RNA was combined with the probe according to the principle of complementary base pairing, resulting in a decrease of the photocurrent. The results show that the absorbance spectrum of visible light is enhanced for CSZFRs compared with pure ZnO. Under visible light irradiation, the photocurrent of CSZFRs is up to 0.1 mA, which can improve the sensitivity of the photoelectrochemical (PEC) biosensor. In the measurement range of 0⁻5.10 nM, the measured concentrations (c) have a good linear relationship with the output photocurrent of the biosensor. The linear regression equation is expressed as I = 0.03256 - 0.0033c (R² = 0.99, S/N = 3) with a detection limit of 0.50 nM. Therefore, this work realizes a rapid and sensitive method for the detection of NV RNA.

[Spectral Analysis of CdZnSe Ternary Quantum Dots Sensitized TiO2 Tubes and Its Application in Visible-Light Photocatalysis].

In this work, cadmium nitrate hexahydrate [Cd(NO3)2 · 6H2O] is as a source of cadmium, zinc nitrate [Zn(NO3)2] as a source of zinc source, and NaHSe as a source of selenium which was prepared through reducing the elemental selenium with sodium borohydride (NaBH4). Then water-soluble Cd1₋xZnxSe ternary quantum dots with different component were prepared by colloid chemistry. The as-prepared Cd1₋xZnx Se ternary quantum dots exhibit stable fluorescent property in aqueous solution, and can still maintain good dispersivity at room temperature for four months. Powder X-ray diffraction (XRD) and high resolution transmission electron microscope (HRTEM) were used to analyze crystal structure and morphology of the prepared Cd1₋xZnxSe. It is found that the as-prepared ternary quantum dots are cubic phase, show as sphere, and the average of particle size is approximate 4 nm. The spectral properties and energy band structure of the as-prepared ternary quantum dots were modulated through changing the atom ratio of elements Zn and Cd. Compared with binary quantum dots CdSe and ZnSe, the ultraviolet-visible (UV-Visible) absorption spectrum and fluorescence (FL) emission spectrum of ternary quantum dots are both red-shift. The composites (Cd0.5 Zn0.5 Se@TNTs) of Cd0.5 Zn0.5 Se ternary quantum dots and TiO2 nanotubes (TNTs) were prepared by directly immerging TNTs into quantum dots dispersive solution for 5 hours. TEM image shows that the Cd0.5 Zn0.5 Se ternary quantum dots were closely combined to nanotube surface. The infrared spectra show that the Ti-Se bond was formed between Cd0.5 Zn0.5 Se ternary quantum dots and TiO2 nanotubes, which improve the stability of the composite. Compared to pristine TNTs, UV-Visible absorption spectrum of the composites is significantly enhanced in the visible region of light. And the absorption band edge of Cd0.5Zn0.5 Se@TNTs red-shift from 400 to 700 nm. The recombination of the photogenerated electron-hole pairs was restrained with the as-prepared ternary quantum dots. Therefore, the visible-light photocatalytic efficiency was greatly improved. After visible-light irradiation for 60 min, the degradation of Cd0.5 Zn0.5 Se@TNTs photocatalysts for RhB is nearly 100%, which is about 3. 3 times of that of pristine TNTs and 2. 5 times of that of pure Cd0.5 Zn0.5 Se ternary quantum dots, respectively.

Fast Learning of Signed Distance Functions From Noisy Point Clouds Via Noise to Noise Mapping.

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. Our novelty lies in the noise to noise mapping which can infer a highly accurate SDF of a single object or scene from its multiple or even single noisy observations. We achieve this by a novel loss which enables statistical reasoning on point clouds and maintains geometric consistency although point clouds are irregular, unordered and have no point correspondence among noisy observations. To accelerate training, we use multi-resolution hash encodings implemented in CUDA in our framework, which reduces our training time by a factor of ten, achieving convergence within one minute. We further introduce a novel schema to improve multi-view reconstruction by estimating SDFs as a prior. Our evaluations under widely-used benchmarks demonstrate our superiority over the state-of-the-art methods in surface reconstruction from point clouds or multi-view images, point cloud denoising and upsampling.

Learning Signed Hyper Surfaces for Oriented Point Cloud Normal Estimation.

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., unoriented normal estimation and normal orientation, and each step is implemented by a separate algorithm. However, previous methods are sensitive to parameter settings, resulting in poor results from point clouds with noise, density variations and complex geometries. In this work, we introduce signed hyper surfaces (SHS), which are parameterized by multi-layer perceptron (MLP) layers, to learn to estimate oriented normals from point clouds in an end-to-end manner. The signed hyper surfaces are implicitly learned in a high-dimensional feature space where the local and global information is aggregated. Specifically, we introduce a patch encoding module and a shape encoding module to encode a 3D point cloud into a local latent code and a global latent code, respectively. Then, an attention-weighted normal prediction module is proposed as a decoder, which takes the local and global latent codes as input to predict oriented normals. Experimental results show that our algorithm outperforms the state-of-the-art methods in both unoriented and oriented normal estimation.

CAP-UDF: Learning Unsigned Distance Functions Progressively from Raw Point Clouds with Consistency-Aware Field Optimization.

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. However, the learned UDF is hard to provide smooth distance fields due to the discontinuous character of point clouds. In this paper, we propose CAP-UDF, a novel method to learn consistency-aware UDF from raw point clouds. We achieve this by learning to move queries onto the surface with a field consistency constraint, where we also enable to progressively estimate a more accurate surface. Specifically, we train a neural network to gradually infer the relationship between queries and the approximated surface by searching for the moving target of queries in a dynamic way. Meanwhile, we introduce a polygonization algorithm to extract surfaces using the gradients of the learned UDF. We conduct comprehensive experiments in surface reconstruction for point clouds, real scans or depth maps, and further explore our performance in unsupervised point normal estimation, which demonstrate non-trivial improvements of CAP-UDF over the state-of-the-art methods.

Synthesis of carbon dot based Schiff bases and selective anticancer activity in glioma cells.

Schiff bases have remarkable anticancer activity and are used for glioma therapy. However, the poor water solubility/dispersibility limits their therapeutic potential in biological systems. To address this issue, carbon dots (CDs) have been utilized to enhance the dispersibility in water and biological efficacy of Schiff bases. The amino groups on the surface of CDs were conjugated effectively with the aldehyde group of terephthalaldehyde to form novel CD-based Schiff bases (CDSBs). The results of the MTT assays demonstrate that CDSBs have significant anticancer activity in glioma GL261 cells and U251 cells, with IC50 values of 17.9 µg mL-1 and 14.9 µg mL-1, respectively. CDSBs have also been found to have good biocompatibility with normal glial cells. The production of reactive oxygen species (ROS) in GL261 glioma cells showed that CDSBs, at a concentration of 44 µg mL-1, resulted in approximately 13 times higher intracellular ROS production than in the control group. These experiments offer evidence that CDSBs induce mitochondrial damage, leading to a reduction in mitochondrial membrane potential in GL261 cells. In particular, in this work, CDs serve not as carriers, but as an integral part of the anticancer drugs, which can expand the role of CDs in cancer treatment.

PMP-Net++: Point Cloud Completion by Transformer-Enhanced Multi-Step Point Moving Paths.

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. We address this problem by formulating completion as point cloud deformation process. Specifically, we design a novel neural network, named PMP-Net++, to mimic behavior of an earth mover. It moves each point of incomplete input to obtain a complete point cloud, where total distance of point moving paths (PMPs) should be the shortest. Therefore, PMP-Net++ predicts unique PMP for each point according to constraint of point moving distances. The network learns a strict and unique correspondence on point-level, and thus improves quality of predicted complete shape. Moreover, since moving points heavily relies on per-point features learned by network, we further introduce a transformer-enhanced representation learning network, which significantly improves completion performance of PMP-Net++. We conduct comprehensive experiments in shape completion, and further explore application on point cloud up-sampling, which demonstrate non-trivial improvement of PMP-Net++ over state-of-the-art point cloud completion/up-sampling methods.

Snowflake Point Deconvolution for Point Cloud Completion and Generation With Skip-Transformer.

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. Our insight into the detailed geometry is to introduce a skip-transformer in the SPD to learn the point splitting patterns that can best fit the local regions. The skip-transformer leverages attention mechanism to summarize the splitting patterns used in the previous SPD layer to produce the splitting in the current layer. The locally compact and structured point clouds generated by SPD precisely reveal the structural characteristics of the 3D shape in local patches, which enables us to predict highly detailed geometries. Moreover, since SPD is a general operation that is not limited to completion, we explore its applications in other generative tasks, including point cloud auto-encoding, generation, single image reconstruction, and upsampling. Our experimental results outperform state-of-the-art methods under widely used benchmarks.

In or on, a study of the influence of the binding site for TiO2 and MIL-101(Cr).

In this work, TiO2 was formed in situ in the internal pores and on the surface of MIL-101(Cr). Density functional theory (DFT) calculations demonstrate that the difference in the binding sites of TiO2 can be attributed to the different solvents used. The two composites were used to photodegrade methyl orange (MO), and the photocatalytic efficiency of TiO2-in-MIL-101(Cr) (90.1% in 120 min) was much stronger than that of TiO2-on-MIL-101(Cr) (14% in 120 min). This is the first work to study the influence of the binding site of TiO2 and MIL-101(Cr). The results show that MIL-101(Cr) modification with TiO2 can promote electron-hole separation, and TiO2-in-MIL-101(Cr) has better performance. Interestingly, the two prepared composites have distinct electron transfer processes. For TiO2-on-MIL-101(Cr), radical trapping and electron paramagnetic resonance (EPR) studies show that O2˙- is the main reactive oxygen species. Based on its band structure, it can be concluded that the electron transfer process of TiO2-on-MIL-101(Cr) conforms to that of a type II heterojunction. However, for TiO2-in-MIL-101(Cr), the EPR and DFT results show that 1O2 is the active substance that is formed from O2 through energy transfer. Therefore, the influence of binding sites should be considered for the improvement of MOF materials.

Fast Learning Radiance Fields by Shooting Much Fewer Rays.

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. To resolve this issue, we introduce a general strategy to speed up the learning procedure for almost all radiance fields based methods. Our key idea is to reduce the redundancy by shooting much fewer rays in the multi-view volume rendering procedure which is the base for almost all radiance fields based methods. We find that shooting rays at pixels with dramatic color change not only significantly reduces the training burden but also barely affects the accuracy of the learned radiance fields. In addition, we also adaptively subdivide each view into a quadtree according to the average rendering error in each node in the tree, which makes us dynamically shoot more rays in more complex regions with larger rendering error. We evaluate our method with different radiance fields based methods under the widely used benchmarks. Experimental results show that our method achieves comparable accuracy to the state-of-the-art with much faster training.

SPU-Net: Self-Supervised Point Cloud Upsampling by Coarse-to-Fine Reconstruction With Self-Projection Optimization.

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. To address this problem, we propose a self-supervised point cloud upsampling network, named SPU-Net, to capture the inherent upsampling patterns of points lying on the underlying object surface. Specifically, we propose a coarse-to-fine reconstruction framework, which contains two main components: point feature extraction and point feature expansion, respectively. In the point feature extraction, we integrate the self-attention module with the graph convolution network (GCN) to capture context information inside and among local regions simultaneously. In the point feature expansion, we introduce a hierarchically learnable folding strategy to generate upsampled point sets with learnable 2D grids. Moreover, to further optimize the noisy points in the generated point sets, we propose a novel self-projection optimization associated with uniform and reconstruction terms as a joint loss to facilitate the self-supervised point cloud upsampling. We conduct various experiments on both synthetic and real-scanned datasets, and the results demonstrate that we achieve comparable performances to state-of-the-art supervised methods.

TiO2-In-MIL-101(Cr) with Visible Light-Induced Peroxidase Activity for Colorimetric Detection of Blood Glucose.

In this work, metal-organic framework MIL-101(Cr) with regular morphology, stable structure, and good dispersion was prepared by the hydrothermal method. MIL-101(Cr) has two different sizes of pores, but after TiO2 nanoparticles (NPs) were in situ prepared, the two pores disappear. The result demonstrates that TiO2 NPs were located in the pores of MIL-101(Cr). TiO2-decorated MIL-101(Cr) forms an inside type II heterojunction and the band gap energy is narrowed, which can promote electron-hole separation and enhance the light absorption. Therefore, the heterojunction shows a high visible light-induced peroxidase-like activity. Kinetic studies exhibit that the K m value of TiO2-in-MIL-101(Cr) to TMB is 0.17 mM, and the affinity of TiO2-in-MIL-101(Cr) is higher than that of natural horseradish peroxidase (HRP). Then, a "turn-on" colorimetric assay based on TiO2-in-MIL-101(Cr) was constructed for the detection of blood glucose. The detection range is 1-100 µM (R 2 = 0.9950) with a limit of detection (LOD) of 1.17 µM. Compared with the clinical method, the constructed colorimetric method has accurate and reliable results for the clinical detection. The anti-interference experiment confirms that the method has high selectivity to glucose.

Electrochemical Detection of Alpha-Fetoprotein Based on Black Phosphorus Nanosheets Modification with Iron Ions.

Black phosphorus nanosheets (BPNSs) were synthesized with liquid exfoliation combined with the ultrasonic method and loaded with Fe3+ by simply mixing. The morphology, structure and electrochemical properties of the synthesized Fe3+/BPNSs were characterized by transmission electron microscopy (TEM), atomic force microscopy (AFM), Raman spectroscopy, X-ray photoelectron spectroscopy (XPS) and cyclic voltammetry (CV), etc. The load of Fe3+ can improve the electrochemical performance of BPNSs and enhance the sensitivity of the detection. Additionally, Fe3+/BPNSs display good biocompatibility. In this study, immunosensors based on Fe3+/BPNSs were constructed to detect alpha-fetoprotein (AFP). The detection is due to the specific binding between the AFP antigen and antibody on the surface of the immunosensors, which can reduce the current response of Fe3+/BPNSs. The immunosensors have a good linear relationship in the range of 0.005 ng·mL-1 to 50 ng·mL-1, and the detection limit is 1.2 pg·mL-1. The results show that surface modification with metal ions is a simple and effective way to improve the electrochemical properties of BPNSs, which will broaden the prospects for the future application of BPNSs in the electrochemical field.

Dual-mode turn-on ratiometric fluorescence sensor based on carbon dots and CuInS2/ZnS quantum dots for detection of chlorotetracycline.

A new ratiometric fluorescence sensor is prepared for selective detection of chlorotetracycline (CTC) through dual-mode fluorescence method. The sensor is composed of carbon dots (CDs) with blue emission and carboxyl-modified CuInS2/ZnS quantum dots (QDs) with dark-red emission. Usually QDs are used as fluorescent probes or signal sources, but it is interesting in this strategy that CuInS2/ZnS QDs innovatively work as quenching agent to reduce the fluorescence of CDs, mainly due to the fluorescence resonance energy transfer (FRET). After the addition of CTC, the interaction between CDs and CuInS2/ZnS QDs is restrained, resulting in the fluorescence recovery of CDs, whilstthe QDs' fluorescence remains unaffected. In this work, CTC is detected in the range of 0-50 µM by conventional fluorescence and synchronous fluorescence methods under an excitation wavelength of 360 nm or Δλ = 90 nm, and the detection limits of the two methods are 0.46 µM and 0.36 µM, respectively. The designed sensor displays good selectivity compared with other tetracycline drugs with similar structure to CTC, different ions and various natural - amino acids. And the sensor can also be applied to determine CTC in tap water and milk.

Construction of Photoelectrochemical DNA Biosensors Based on TiO2@Carbon Dots@Black Phosphorous Quantum Dots.

In this work, carbon dots (CDs) and black phosphorus quantum dots (BPQDs) were used to decorate titanium dioxide to enhance the photoelectrochemical (PEC) properties of the nanocomposites (TiO2@CDs@BPQDs), and the modified nanocomposites were used to sensitively detect DNA. We used the hydrothermal method and citric acid as a raw material to prepare CDs with good dispersion and strong fluorescence properties. BPQDs with a uniform particle size were prepared from black phosphorus crystals. The nanocomposites were characterized by fluorescence spectroscopy, UV-Vis absorption spectroscopy, Fourier transform infrared spectroscopy (FTIR) and transmission electron microscopy (TEM). The preparation method of the working electrode was explored, the detection conditions were optimized, and the sensitive detection of target DNA was achieved. The results demonstrate that CDs and BPQDs with good optical properties were successfully prepared, and they were successfully combined with TiO2 to improve the PEC performance of TiO2@CDs@BPQDs. The TiO2-based PEC DNA detection method was constructed with a detection limit of 8.39 nM. The constructed detection method has many advantages, including good sensitivity, a wide detection range, and good specificity. This work provides a promising PEC strategy for the detection of other biomolecules.

Fine-Grained 3D Shape Classification With Hierarchical Part-View Attention.

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. Each category consists of several subcategories at a fine-grained level. According to our experiments under this fine-grained dataset, we find that state-of-the-art methods are significantly limited by the small variance among subcategories in the same category. To resolve this problem, we further propose a novel fine-grained 3D shape classification method named FG3D-Net to capture the fine-grained local details of 3D shapes from multiple rendered views. Specifically, we first train a Region Proposal Network (RPN) to detect the generally semantic parts inside multiple views under the benchmark of generally semantic part detection. Then, we design a hierarchical part-view attention aggregation module to learn a global shape representation by aggregating generally semantic part features, which preserves the local details of 3D shapes. The part-view attention module hierarchically leverages part-level and view-level attention to increase the discriminability of our features. The part-level attention highlights the important parts in each view while the view-level attention highlights the discriminative views among all the views of the same object. In addition, we integrate a Recurrent Neural Network (RNN) to capture the spatial relationships among sequential views from different viewpoints. Our results under the fine-grained 3D shape dataset show that our method outperforms other state-of-the-art methods. The FG3D dataset is available at https://github.com/liuxinhai/FG3D-Net.

Active Arrangement of Small Objects in 3D Indoor Scenes.

Small object arrangement is very important for creating detailed and realistic 3D indoor scenes. In this article, we present an interactive framework based on active learning to help users create customized arrangements for small objects according to their preferences. To achieve this with minimal user effort, we first learn the prior knowledge about small object arrangement from a 3D indoor scene dataset through a probability mining method, which forms the initial guidance for arranging small objects. Then, users are able to express their preferences on a few small object categories, which are automatically propagated to all the other categories via a novel active learning approach. In the propagation process, we introduce a novel metric to obtain the propagation weights, which measures the degree of interchangeability between two small object categories, and is calculated based on a spatial embedding model learned from the small object neighborhood information extracted from the 3D indoor scene dataset. Experiments show that our framework is able to help users effectively create customized small object arrangements with little effort.