Pedestrian-Accessible Infrastructure Inventory: Enabling and Assessing Zero-Shot Segmentation on Multi-Mode Geospatial Data for All Pedestrian Types.

In this paper, a Segment Anything Model (SAM)-based pedestrian infrastructure segmentation workflow is designed and optimized, which is capable of efficiently processing multi-sourced geospatial data, including LiDAR data and satellite imagery data. We used an expanded definition of pedestrian infrastructure inventory, which goes beyond the traditional transportation elements to include street furniture objects that are important for accessibility but are often omitted from the traditional definition. Our contributions lie in producing the necessary knowledge to answer the following three questions. First, how can mobile LiDAR technology be leveraged to produce comprehensive pedestrian-accessible infrastructure inventory? Second, which data representation can facilitate zero-shot segmentation of infrastructure objects with SAM? Third, how well does the SAM-based method perform on segmenting pedestrian infrastructure objects? Our proposed method is designed to efficiently create pedestrian-accessible infrastructure inventory through the zero-shot segmentation of multi-sourced geospatial datasets. Through addressing three research questions, we show how the multi-mode data should be prepared, what data representation works best for what asset features, and how SAM performs on these data presentations. Our findings indicate that street-view images generated from mobile LiDAR point-cloud data, when paired with satellite imagery data, can work efficiently with SAM to create a scalable pedestrian infrastructure inventory approach with immediate benefits to GIS professionals, city managers, transportation owners, and walkers, especially those with travel-limiting disabilities, such as individuals who are blind, have low vision, or experience mobility disabilities.

Robust Face Alignment via Inherent Relation Learning and Uncertainty Estimation.

Human tends to locate the facial landmarks with heavy occlusion by their relative position to the easily identified landmarks. The clue is defined as the landmark inherent relation while it is ignored by most existing methods. In this paper, we present Dynamic Sparse Local Patch Transformer (DSLPT), a novel face alignment framework for the inherent relation learning and uncertainty estimation. Unlike most existing methods that regress facial landmarks directly from global features, the DSLPT first generates a rough representation of each landmark from a local patch cropped from the feature map and then adaptively aggregates them by a case dependent inherent relation. Finally, the DSLPT predicts the coordinate and uncertainty of each landmark by regressing their probability distribution from the output features. Moreover, we introduce a coarse-to-fine framework to incorporate with DSLPT for an improved result. In the framework, the position and size of each patch are determined by the probability distribution of the corresponding landmark predicted in the previous stage. The dynamic patches will ensure a fine-grained landmark representation for inherent relation learning so that a rough prediction result can gradually converge to the target facial landmarks. We integrate the coarse-to-fine model into an end-to-end training pipeline and carry out experiments on the mainstream benchmarks. The results demonstrate that the DSLPT achieves state-of-the-art performance with much less computational complexity. The codes and models are available at https://github.com/Jiahao-UTS/DSLPT.

Learning Graph Representations Through Learning and Propagating Edge Features.

Graph convolutional networks have achieved considerable success in various graph domain tasks. Recently, numerous types of graph convolutional networks have been developed. A typical rule for learning a node's feature in these graph convolutional networks is to aggregate node features from the node's local neighborhood. However, in these models, the interrelation information between adjacent nodes is not well-considered. This information could be helpful to learn improved node embeddings. In this article, we present a graph representation learning framework that generates node embeddings through learning and propagating edge features. Instead of aggregating node features from a local neighborhood, we learn a feature for each edge and update a node's representation by aggregating local edge features. The edge feature is learned from the concatenation of the edge's starting node feature, the input edge feature, and the edge's end node feature. Unlike node feature propagation-based graph networks, our model propagates different features from a node to its neighbors. In addition, we learn an attention vector for each edge in aggregation, enabling the model to focus on important information in each feature dimension. By learning and aggregating edge features, the interrelation between a node and its neighboring nodes is integrated in the aggregated feature, which helps learn improved node embeddings in graph representation learning. Our model is evaluated on graph classification, node classification, graph regression, and multitask binary graph classification on eight popular datasets. The experimental results demonstrate that our model achieves improved performance compared with a wide variety of baseline models.

Molecular homogeneity of GB1 revealed by single molecule force spectroscopy.

In single molecule studies, the ergodic hypothesis is inherently assumed, which states that the time average of a physical quantity of a single member of an ensemble is the same as the average of the same quantity on the whole ensemble at a given time. This hypothesis implies the homogeneity of a molecular ensemble of a system of interest. However, it is difficult to test the validity of the ergodic hypothesis experimentally. Recent theoretical work suggested that heterogeneity may be widely present in single molecule force spectroscopy studies. Here we used atomic force microscope based single molecule force spectroscopy to examine the molecular homogeneity/heterogeneity of a small globular protein GB1 in its mechanical unfolding reaction. Using a polyprotein (GB1)4, we directly measured the ensemble average and time average for a single molecule of the mechanical unfolding force and kinetic parameters that characterize the mechanical unfolding free energy profile of GB1. Our results showed that the ensemble averages of these physical quantities are indeed the same as the time averages for single molecules, and individual molecules did not show any differences amongst them in these physical quantities. These results are consistent with the expectation of the ergodic hypothesis and indicate that GB1 is a homogeneous molecular ensemble in its mechanical unfolding reaction on the time scale of our force spectroscopy experiments.

When Dynamic Diselenide Bonds Meet Dynamic Imine Bonds in Polymeric Materials.

In both natural and artificial functional systems, the cooperation between different dynamic interactions is of vital importance for realizing complicated functions. Dynamic covalent bonds are one kind of relatively stable dynamic interactions and have shown synergistic effect in natural systems such as functional proteins. However, synergistic interactions between different dynamic covalent bonds in polymeric materials are still unclear. Herein, polymeric materials containing diselenide and imine bonds are prepared, and then the synergistic effect between the two dynamic covalent bonds is quantitatively evaluated in typical processes of dynamic materials. The results reveal that dynamic covalent bonds show weak synergistic effect in the degradation process and have strong synergistic effect in stress relaxation process. Therefore, introducing multiple dynamic covalent bonds in polymeric materials can extensively enhance their dynamic properties.

Laser-Induced Remote Healing of Stretchable Diselenide-Containing Conductive Composites.

Remotely controlled on-demand functional healing is vital to components that are difficult to access and repair in distance such as satellites and unmanned cruising aircrafts. Compared with other stimuli, a blue laser is a better choice to input energy to the damaged area in distance because of its high energy density and low dissipation through the air. Herein, diselenide-containing polyurethane (PUSe) is first employed to fabricate visible light-responsive stretchable conductive composites with multiwalled carbon nanotubes (MWCNTs). Then, laser-induced remote healing was realized based on the characteristics of long-distance propagation of lasers and the dynamic properties of diselenide bonds. Moreover, the PUSe/MWCNT composite film can be used to transfer an electrical signal in the circuit containing a signal generator. This laser-induced remote healing of conductivity paves the way for developing healing conductors which are difficult to access and repair.

Single-Molecule Force Spectroscopy Studies of Missense Titin Mutations That Are Likely Causing Cardiomyopathy.

The giant muscle protein titin plays important roles in heart function. Mutations in titin have emerged as a major cause of familial cardiomyopathy. Missense mutations have been identified in cardiomyopathy patients; however, it is challenging to distinguish disease-causing mutations from benign ones. Given the importance of titin mechanics in heart function, it is critically important to elucidate the mechano-phenotypes of cardiomyopathy-causing mutations found in the elastic I-band part of cardiac titin. Using single-molecule atomic force microscopy (AFM) and equilibrium chemical denaturation, we investigated the mechanical and thermodynamic effects of two missense mutations, R57C-I94 and S22P-I84, found in the elastic I-band part of cardiac titin that were predicted to be likely causing cardiomyopathy by bioinformatics analysis. Our AFM results showed that mutation R57C had a significant destabilization effect on the I94 module. R57C reduced the mechanical unfolding force of I94 by ∼30-40 pN, accelerated the unfolding kinetics, and decelerated the folding. These effects collectively increased the unfolding propensity of I94, likely resulting in altered titin elasticity. In comparison, S22P led to only modest destabilization of I84, with a decrease in unfolding force by ∼10 pN. It is unlikely that such a modest destabilization would lead to a change in titin elasticity. These results will serve as the first step toward elucidating mechano-phenotypes of cardiomyopathy-causing mutations in the elastic I-band.

Parking Slot Detection on Around-View Images Using DCNN.

Due to the complex visual environment and incomplete display of parking slots on around-view images, vision-based parking slot detection is a major challenge. Previous studies in this field mostly use the existing models to solve the problem, the steps of which are cumbersome. In this paper, we propose a parking slot detection method that uses directional entrance line regression and classification based on a deep convolutional neural network (DCNN) to make it robust and simple. For parking slots with different shapes and observed from different angles, we represent the parking slot as a directional entrance line. Subsequently, we design a DCNN detector to simultaneously obtain the type, position, length, and direction of the entrance line. After that, the complete parking slot can be easily inferred using the detection results and prior geometric information. To verify our method, we conduct experiments on the public ps2.0 dataset and self-annotated parking slot dataset with 2,135 images. The results show that our method not only outperforms state-of-the-art competitors with a precision rate of 99.68% and a recall rate of 99.41% on the ps2.0 dataset but also performs a satisfying generalization on the self-annotated dataset. Moreover, it achieves a real-time detection speed of 13 ms per frame on Titan Xp. By converting the parking slot into a directional entrance line, the specially designed DCNN detector can quickly and effectively detect various types of parking slots.

Wavelength-Controlled Light-Responsive Polymer Vesicle Based on Se-S Dynamic Chemistry.

Wavelength-controlled Se-S dynamic chemistry was put forward recently as a convenient way to regulate the balance of a selenide sulfide exchange reaction. In this paper, we synthesized an asymmetric polymeric amphiphile linked with a Se-S bond and then induced it to self-assemble into vesicles in water. When the visible light was applied to the assembly solution with addition of toluene, Se-S bonds containing vesicles were ruptured. Thus, the wavelength-controlled light responses of relatively stable polymer assembly were accomplished by introduction of the Se-S dynamic covalent bond, and the response mechanism of the Se-S bond in the vesicle was explored by nuclear magnetic resonance (NMR), gel permeation chromatography (GPC), and X-ray photoelectron spectroscopy (XPS). The results indicated that fracture of the Se-S bond led to the dissociation of assembly. Introduction of Se-S dynamic chemistry into the molecular assembly area enriched the light-responsive polymer systems and would bring many potential applications in the future.

Single molecule force spectroscopy reveals that the oxidation state of cobalt ions plays an important role in enhancing the mechanical stability of proteins.

Engineered bi-histidine (biHis)-based metal chelation is a general and robust method to enhance the mechanical stability of proteins. Here we used single molecule force spectroscopy techniques to investigate the effect of binding of Co2+/Co3+ on the mechanical stability of an engineered biHis mutant of protein GB1, G6-53. We found that the binding of Co2+/Co3+ can lead to an enhancement of the mechanical stability of G6-53, but the degree of enhancement is drastically different. The binding of Co2+ can only lead to marginal enhancement of G6-53's mechanical stability, while Co3+ has a much stronger effect. This large difference is likely due to the large difference in thermodynamic stability and kinetic lability of Co2+ and Co3+ complexes. These results opened up new avenues towards fine tuning the mechanical properties of proteins.

Visible-light-induced metathesis reaction between diselenide and ditelluride.

The Se-Te bond was prepared by a metathesis reaction between diselenides and ditellurides, which could be manipulated with the presence or absence of visible light. Additionally, the apparent activation energy of the exchange process was measured to be only 28.01 kJ mol-1, explaining the high reactivity and sensitivity to light.

Diselenide-Containing Polymeric Vesicles with Osmotic Pressure Response.

Mechanophore is a kind of functional group that can undergo chemical reactions when given mechanical force stimuli. In this paper, osmotic pressure was used as an external force to trigger a diselenide exchange reaction. A diselenide bond containing block polymer capable of self-assembling to a vesicle structure and an ester bond containing a counterpart were synthesized in this study. When NaCl was added into the solution to generate the osmotic pressure difference inside and outside vesicles, diselenide containing vesicles were ruptured, while the ester bond counterpart stayed still. Further investigation into the chemical composition of both vesicles indicated the occurrence of the diselenide exchange reaction. The osmotic pressure response of the diselenide bond enriched the diselenide dynamic covalent chemistry and offers a potential application in a controlled release system.

Surface Modification Based on Diselenide Dynamic Chemistry: Towards Liquid Motion and Surface Bioconjugation.

Surface modification is an important technique in fields, such as, self-cleaning, surface patterning, sensing, and detection. The diselenide bond was shown to be a dynamic covalent bond that can undergo a diselenide metathesis reaction simply under visible light irradiation. Herein we develop this diselenide dynamic chemistry into a versatile surface modification method with a fast response and reversibility. The diselenide bond could be modified onto various substrates, such as, PDMS, quartz, and ITO conductive film glass. Different functional diselenide molecules could then be immobilized onto the surface via diselenide metathesis reaction. We demonstrated that by using this modification method we could achieve liquid motion in a capillary tube under light illumination. We also show that this approach has the potential to serve as an efficient modification method for surface bioconjugation, which has practical applications in clinical usage.

Selenium-Functionalized Graphene Oxide That Can Modulate the Balance of Reactive Oxygen Species.

Graphene oxide (GO) is an important two-dimensional material since it is water-soluble and can be functionalized to adapt to different applications. However, the current covalent functionalization methods usually require hash conditions, long duration, and sometimes even multiple steps, while noncovalent functionalization is inevitably unstable, especially under a physiological environment where competing species exist. Diselenide bond is a dynamic covalent bond and can respond to both redox conditions and visible light irradiation in a sensitive manner. Thus, in this work by combining the stimuli response of diselenide bond and the oxidative/radical attackable nature of GO, we achieved the in situ covalent functionalization of GO simply by stirring GO with diselenide-containing molecules in aqueous solution. The covalent functionalization was proved by Fourier transform infrared, time-of-flight secondary ion mass spectrometry, atomic force microscopy, thermogravimetric analysis, and so forth, and the functionalization mechanism was deduced to involve both redox reaction and radical addition reaction according to the X-ray photoelectron spectrscopy, atomic emission spectroscopy, and Raman spectroscopy. Moreover, we modified GO with a biocompatible diselenide-containing polymer (mPEGSe)2 and found selenium-functionalized GO could modulate the balance of reactive oxygen species (ROS). GOSe could decrease ROS level by accelerating the reduction of peroxides when the ROS concentration is high while boosting the ROS level by in situ generating ROS when its concentration is relatively low.

Selenium-Doped Carbon Quantum Dots for Free-Radical Scavenging.

Heteroatom doping is an effective way to adjust the fluorescent properties of carbon quantum dots. However, selenium-doped carbon dots have rarely been reported, even though selenium has unique chemical properties such as redox-responsive properties owing to its special electronegativity. Herein, a facile and high-output strategy to fabricate selenium-doped carbon quantum dots (Se-CQDs) with green fluorescence (quantum yield 7.6 %) is developed through the hydrothermal treatment of selenocystine under mild conditions. Selenium heteroatoms endow the Se-CQDs with redox-dependent reversible fluorescence. Furthermore, free radicals such as . OH can be effectively scavenged by the Se-CQDs. Once Se-CQDs are internalized into cells, harmful high levels of reactive oxygen species (ROS) in the cells are decreased. This property makes the Se-CQDs capable of protecting biosystems from oxidative stress.

Dynamic Chemistry of Selenium: Se-N and Se-Se Dynamic Covalent Bonds in Polymeric Systems.

The application of selenium in the responsive polymer system and the enzyme mimic system have been well studied. Our group initiated this line of research in 2009 by first extending selenium chemistry to dynamic chemistry. In this article, the discovery, progress, and application of selenium-related dynamic covalent bonds will be introduced. The dynamic property of Se-N bond and Se-Se bond were revealed and have been applied in the polymer system as enzyme mimic and self-healing materials, respectively. Further studies that need to be done and potential application of selenium-related dynamic chemistry will also be discussed.