Fast Building Instance Proxy Reconstruction for Large Urban Scenes.

Digitalization of large-scale urban scenes (in particular buildings) has been a long-standing open problem, which attributes to the challenges in data acquisition, such as incomplete scene coverage, lack of semantics, low efficiency, and low reliability in path planning. In this paper, we address these challenges in urban building reconstruction from aerial images, and we propose an effective workflow and a few novel algorithms for efficient 3D building instance proxy reconstruction for large urban scenes. Specifically, we propose a novel learning-based approach to instance segmentation of urban buildings from aerial images followed by a voting-based algorithm to fuse the multi-view instance information to a sparse point cloud (reconstructed using a standard Structure from Motion pipeline). Our method enables effective instance segmentation of the building instances from the point cloud. We also introduce a layer-based surface reconstruction method dedicated to the 3D reconstruction of building proxies from extremely sparse point clouds. Extensive experiments on both synthetic and real-world aerial images of large urban scenes have demonstrated the effectiveness of our approach. The generated scene proxy models can already provide a promising 3D surface representation of the buildings in large urban scenes, and when applied to aerial path planning, the instance-enhanced building proxy models can significantly improve data completeness and accuracy, yielding highly detailed 3D building models.

PathNet: Path-Selective Point Cloud Denoising.

Current point cloud denoising (PCD) models optimize single networks, trying to make their parameters adaptive to each point in a large pool of point clouds. Such a denoising network paradigm neglects that different points are often corrupted by different levels of noise and they may convey different geometric structures. Thus, the intricacy of both noise and geometry poses side effects including remnant noise, wrongly-smoothed edges, and distorted shape after denoising. We propose PathNet, a path-selective PCD paradigm based on reinforcement learning (RL). Unlike existing efforts, PathNet enables dynamic selection of the most appropriate denoising path for each point, best moving it onto its underlying surface. We have two more contributions besides the proposed framework of path-selective PCD for the first time. First, to leverage geometry expertise and benefit from training data, we propose a noise- and geometry-aware reward function to train the routing agent in RL. Second, the routing agent and the denoising network are trained jointly to avoid under- and over-smoothing. Extensive experiments show promising improvements of PathNet over its competitors, in terms of the effectiveness for removing different levels of noise and preserving multi-scale surface geometries. Furthermore, PathNet generalizes itself more smoothly to real scans than cutting-edge models.

CSDN: Cross-Modal Shape-Transfer Dual-Refinement Network for Point Cloud Completion.

How will you repair a physical object with some missings? You may imagine its original shape from previously captured images, recover its overall (global) but coarse shape first, and then refine its local details. We are motivated to imitate the physical repair procedure to address point cloud completion. To this end, we propose a cross-modal shape-transfer dual-refinement network (termed CSDN), a coarse-to-fine paradigm with images of full-cycle participation, for quality point cloud completion. CSDN mainly consists of "shape fusion" and "dual-refinement" modules to tackle the cross-modal challenge. The first module transfers the intrinsic shape characteristics from single images to guide the geometry generation of the missing regions of point clouds, in which we propose IPAdaIN to embed the global features of both the image and the partial point cloud into completion. The second module refines the coarse output by adjusting the positions of the generated points, where the local refinement unit exploits the geometric relation between the novel and the input points by graph convolution, and the global constraint unit utilizes the input image to fine-tune the generated offset. Different from most existing approaches, CSDN not only explores the complementary information from images but also effectively exploits cross-modal data in the whole coarse-to-fine completion procedure. Experimental results indicate that CSDN performs favorably against twelve competitors on the cross-modal benchmark.

Discriminating the eight genotypes of the porcine circovirus type 2 with TaqMan-based real-time PCR.

BACKGROUND: The porcine circovirus type 2 (PCV2) is divided into eight genotypes including the previously described genotypes PCV2a to PCV2f and the two new genotypes PCV2g and PCV2h. PCV2 genotyping has become an important task in molecular epidemiology and to advance research on the prophylaxis and pathogenesis of PCV2 associated diseases. Standard genotyping of PCV2 is based on the sequencing of the viral genome or at least of the open reading frame 2. Although, the circovirus genome is small, classical sequencing is time consuming, expensive, less sensitive and less compatible with mass testing compared with modern real-time PCR assays. Here we report about a new PCV2 genotyping method using qPCR. METHODS: Based on the analysis of several hundred PCV2 full genome sequences, we identified PCV2 genotype specific sequences or single-nucleotide polymorphisms. We designed six TaqMan PCR assays that are specific for single genotypes PCV2a to PCV2f and two qPCRs targeting two genotypes simultaneously (PCV2g/PCV2d and PCV2h/PCV2c). To improve specific binding of oligonucleotide primers and TaqMan probes, we used locked nucleic acid technology. We evaluated amplification efficiency, diagnostic sensitivity and tested assay specificity for the respective genotypes. RESULTS: All eight PCV2 genotype specific qPCRs demonstrated appropriate amplification efficiencies between 91 and 97%. Testing samples from an epidemiological field study demonstrated a diagnostic sensitivity of the respective genotype specific qPCR that was comparable to a highly sensitive pan-PCV2 qPCR system. Genotype specificity of most qPCRs was excellent. Limited unspecific signals were obtained when a high viral load of PCV2b was tested with qPCRs targeting PCV2d or PCV2g. The same was true for the PCV2a specific qPCR when high copy numbers of PCV2d were tested. The qPCR targeting PCV2h/PCV2c showed some minor cross-reaction with PCV2d, PCV2f and PCV2g. CONCLUSION: Genotyping of PCV2 is important for routine diagnosis as well as for epidemiological studies. The introduced genotyping qPCR system is ideal for mass testing and should be a valuable complement to PCV2 sequencing, especially in the case of simultaneous infections with multiple PCV2 genotypes, subclinically infected animals or research studies that require large sample numbers.

Isolation and characterization of endothelial progenitor cells from canine bone marrow.

Endothelial progenitor cells (EPC) are located predominantly in the bone marrow. These cells are useful for treating human vascular diseases; they also are a possible target for restricting blood vessel growth for tumors. Little is known about canine EPC. We investigated a bone marrow EPC isolation method that combines the whole bone marrow culture method and the differential adherent speed method using stillborn canines. MTT proliferation, flow cytometry detection, Dil-ac-LDL uptake, FITC-UEA-1 binding and matrigel assays were used to identify and characterize EPC. We isolated two types of EPC: early EPC and late EPC. We found that isolated cells produced typical colony and cobblestone morphology, and were positive for CD31, CD34, CD133 and VEGFR-2. Significant differences were observed in the intensity of expression between early and late EPC, which suggests their different roles during angiogenesis and vasculogenesis. Both early and late EPC were positive for Dil-ac-LDL and FITC-UEA-1, and displayed tube formation when re-suspended in matrigel, both of which are important functional criteria for identifying EPC. Our method is a novel, effective and efficient way to produce enriched EPC.

Testosterone propionate can promote effects of acellular nerve allograft-seeded bone marrow mesenchymal stem cells on repairing canine sciatic nerve.

Peripheral human nerves fail to regenerate across long tube implants (>2 cm), and tissue-engineered nerve grafts represent a promising treatment alternative. The present study aims to investigate the testosterone propionate (TP) repair effect of acellular nerve allograft (ANA) seeded with allogeneic bone marrow mesenchymal stem cells (BMSCs) on 3-cm canine sciatic nerve defect. ANA cellularized with allogeneic BMSCs was implanted to the defect, and TP was injected into the lateral crus of the defected leg. The normal group, the autograft group, the ANA + BMSCs group, the ANA group, and the nongrafted group were used as control. Five months postoperatively, dogs in the TP + ANA + BMSCs group were capable of load bearing, normal walking, and skipping, the autograft group and the ANA + BMSCs group demonstrated nearly the same despite a slight limp. The compound muscle action potentials (CMAPs) on the injured side to the uninjured site in the TP + ANA + BMSCs group were significantly higher than that in the ANA + BMSCs group [CMAPs ratio at A: F(3, 20) = 191.40; 0.02, CMAPs ratio at B: F(3, 20) = 43.27; 0.01]. Masson trichrome staining revealed that in the TP + ANA + BMSCs group, both the diameter ratio of the myelinated nerve and the thickness ratio of regenerated myelin sheath were significantly larger than that in the other groups [the diameter of myelinated nerve fibers: F(3, 56) = 13.45; P < .01, the thickness ratio of regenerated myelin sheath: F(3, 56) = 51.25; P < .01]. In conclusion, TP could significantly increase the repairing effects of the ANA + BMSCs group, and their combination was able to repair 3-cm canine sciatic nerve defect. It therefore represents a promising therapeutic approach.

Trigger factor assisted self-assembly of canine parvovirus VP2 protein into virus-like particles in Escherichia coli with high immunogenicity.

Canine parvovirus (CPV) has been considered to be an important pathogen, which can cause acute infectious disease in canids. Although current vaccines are effective in preventing CPV infection, safety problems still remain unsolved. In this study, a subunit vaccine against CPV based on virus-like particles (VLPs) with good safety and immunogenicity is reported. Soluble CPV VP2 protein was produced by co-expression of chaperone trigger factor (Tf16) in Escherichia coli (E.coli), and assembled into CPV VLPs which could be affected by NaCl and pH. At 250 mM NaCl pH 8.0, the VLPs co-expressed with Tf16 had similar size (25 nm) and shape with the authentic virus capsid under the transmission electron microscopy (TEM), which is also in accordance with the dynamic light scattering (DLS) data. Immunization with these particles could induce high-titer hemagglutination inhibition (1:12288) and neutralizing antibodies (1:6144) in guinea pigs. Splenic cells of them could secrete IFN-γ and IL-4 after stimulation by CPV. Thus, the VLPs produced by the new approach with high yield and immunogenicity could be a potential candidate for CPV vaccine.

Transplantation of Amniotic Scaffold-Seeded Mesenchymal Stem Cells and/or Endothelial Progenitor Cells From Bone Marrow to Efficiently Repair 3-cm Circumferential Urethral Defect in Model Dogs.

The treatment options for patients with a urethral defect are limited by the availability of autologous tissues. We hypothesized that transplantation of decellularized human amniotic scaffolds (dHAS) seeded with allogeneic bone marrow mesenchymal cells (BMSCs) and/or endothelial progenitor cells (EPCs) may serve as a promising repair strategy for long segment of circumferential urethral defect. To verify the hypothesis, with urinary catheterization, a 3-cm segment of whole urethra in 25 male mongrel dogs was excised and replaced by dHAS seeded with allogeneic BMSCs and/or EPCs. Postoperative observation and ascending urethrogram found that dHAS+BMSCs+EPCs and dHAS+EPCs groups demonstrated unhindered urination and capacious urethral caliber, which were similar to the normal group, while urethrostenosis was revealed in dHAS+BMSCs, dHAS, and sham-operated groups, with the shortest narrow section in dHAS+BMSCs group and the longest in sham-operated group. Urethral anatomy check and histological analyses showed that new urethral mucosa composed of stratified columnar epithelium completely covered on the inner surface of the graft site in dHAS+BMSCs+EPCs and dHAS+EPCs groups, but the middle epithelium was thin in dHAS+EPCs group, while incompletely covered in dHAS+BMSCs, dHAS, and sham-operated groups, and there were monolayer epithelial cells at the urethrostenosis in dHAS+BMSCs and dHAS groups. In addition, abundant new vessel and blood sinus showed at submucosa in dHAS+BMSCs+EPCs and dHAS+EPCs groups, instead of the scar tissue of collagen deposition and structural distortion at the urethrostenosis in dHAS+BMSCs, dHAS, and sham-operated groups. This study demonstrates that dHAS seeded with BMSCs+EPCs or EPCs can successfully repair a 3-cm circumferential urethral defect in model dogs, but the former works best. This technology may provide some references for human clinical trials on long segment of circumferential urethral defect repair.

Automatic Constraint Detection for 2D Layout Regularization.

In this paper, we address the problem of constraint detection for layout regularization. The layout we consider is a set of two-dimensional elements where each element is represented by its bounding box. Layout regularization is important in digitizing plans or images, such as floor plans and facade images, and in the improvement of user-created contents, such as architectural drawings and slide layouts. To regularize a layout, we aim to improve the input by detecting and subsequently enforcing alignment, size, and distance constraints between layout elements. Similar to previous work, we formulate layout regularization as a quadratic programming problem. In addition, we propose a novel optimization algorithm that automatically detects constraints. We evaluate the proposed framework using a variety of input layouts from different applications. Our results demonstrate that our method has superior performance to the state of the art.