Search

We study the domain adaptation task for action recognition, namely domain adaptive action recognition, which aims to effectively transfer action recognition power from a label-sufficient source domain to a label-free target domain. Since actions are performed by humans, it is crucial to exploit human cues in videos when recognizing actions across domains. However, existing methods are prone to losing human cues but prefer to exploit the correlation between non-human contexts and associated actions for recognition, and the contexts of interest agnostic to actions would reduce recognition performance in the target domain. To overcome this problem, we focus on uncovering human-centric action cues for domain adaptive action recognition, and our conception is to investigate two aspects of human-centric action cues, namely human cues and human-context interaction cues. Accordingly, our proposed Human-Centric Transformer (HCTransformer) develops a decoupled human-centric learning paradigm to explicitly concentrate on human-centric action cues in domain-variant video feature learning. Our HCTransformer first conducts human-aware temporal modeling by a human encoder, aiming to avoid a loss of human cues during domain-invariant video feature learning. Then, by a Transformer-like architecture, HCTransformer exploits domain-invariant and action-correlated contexts by a context encoder, and further models domain-invariant interaction between humans and action-correlated contexts. We conduct extensive experiments on three benchmarks, namely UCF-HMDB, Kinetics-NecDrone and EPIC-Kitchens-UDA, and the state-of-the-art performance demonstrates the effectiveness of our proposed HCTransformer.

This study aimed to explore the protective effect and potential mechanism of Nostoc communeVauch. polysaccharide (NCVP) on lead (Pb)-poisoning mice. NCVP improved Pb-induced hepatorenal toxicity and inflammatory responses and modulated key indicators of antioxidant capacity. Moreover, the down-regulation of critical proteins of the Nrf2 pathway induced by Pb could be reversed after NCVP intervention. In addition, NCVP maintained the diversity of gut bacteriobiota and restored the relative abundance of f_Prevotellaceae, g_Alloprevotella, and f_ Eubacterium_coprostanoligenes_group reduced by Pb. Also, NCVP regulated the diversity and abundance of gut mycobiota affected by Pb. Specifically, Pb decreased the proportion of pathogenic species (g_Fusarium, p_Basidiomycota, g_Alternaria, g_Aspergillus, and g_Candida) while NCVP increased the abundance of probiotics species (g_Kazachstaniaand p_Ascomycota). Furthermore, the metabolomic analysis found that NCVP significantly altered a range of microbial metabolites, including porphobilinogen, cromakalim, salidroside, and trichostatin A, which has significant associations with specific gut bacteriobiota or mycobiota. These altered metabolites are involved in primary bile acid biosynthesis, metabolism of xenobiotics by cytochrome P450, lysine degradation, and other metabolic pathways. Overall, our findings indicate that NCVP might be an excellent natural product for eliminating Pb-induced hepatorenal toxicity, possibly by regulating gut bacteriome, mycobiome and metabolome.

Accurately estimating the six-degrees of freedom (DoF) pose of unseen objects is crucial for successful robotic manipulation in industrial automation. Some existing methods for this task rely on prior knowledge of individual objects, i.e., the model must be trained on the exact object instance or object category. Others perform unseen object pose estimation but are limited in their feature learning and pose refinement ability. To address these problems, we propose an unseen object pose estimation method that follows a coarse-to-fine framework and leverages the powerful learning ability of diffusion models. We introduce a diffusion model for generating object poses, and conduct a comparison between the generated poses and the original pose to determine the optimal one. We design a novel pose estimation module to provide coarse poses for the PoseDiffusion. This module comprises two feature extraction modules that extract global and masked features. In addition, we propose a strategy to estimate the pose by comparing the similarity between rendered and query poses. The renderings of an unseen object from various viewpoints are generated from its computer-aided design (CAD) model. Our method requires a CAD model of the unseen object only during inference, a scenario well suited to industrial applications. Experimental evaluation on benchmark datasets demonstrates that the proposed framework outperforms existing approaches, achieving state-of-the-art performance in six-DoF object pose estimation.

Estimating the 6-D pose of an object is a vital and challenging task for robot vision systems in industrial robotic grasping. With the wide use of 3-D cameras, the additional acquired depth image provides geometric information of the scene to increase the pose estimation performance but leads to a challenge, fully leveraging the two-modal data, the color image and the depth image. Previous works usually adopt two individual strategies to handle the data, which suffer from limited accuracy and efficiency since the two complementary data are not fully explored. Thus, we propose a depth adaptive feature extraction and dense prediction network that decouples the scale-dependent and the scale-invariant information from the depth image. The former guides the network to perceive the 3-D structure of the scene, and the latter, together with color image, provides the scene textures for feature extraction. The proposed network not only fuses multimodal textures but also retains their 3-D structure. In addition, a dense prediction strategy is adopted to regress the object pose; this approach can mitigate the instability caused by outliers. We conduct various evaluations on a real-world industrial dataset to illustrate the advantages of the proposed approach; and a practical robotic grasping platform is presented to demonstrate its application performance.

The effects of 1 MeV electron irradiation on $\beta $ -Ga2O3 Schottky barrier diodes (SBDs) were studied in this work. After an electron irradiation fluence of $1\times 10\,\,^{{16}}$ cm−2, the forward turn-on voltage of $\beta $ -Ga2O3 SBD decreased from 0.82 to 0.77 V, the reverse current at −100 V decreased by 79%, the height of the Schottky barrier decreased from 1.01 to 0.92 eV, and the ideality factor increased from 1.04 to 1.13, indicating that the electrical characteristics of the $\beta $ -Ga2O3 SBD had degraded. Through ${C}$ – ${V}$ testing, a carrier removal rate of 7.9 cm−1 was determined. The deep-level transient spectroscopy (DLTS) results showed that the concentration of native defects at the ${E} _{C} -0.75$ position increased sharply after irradiation, which may be the main reason for the degradation of the $\beta $ -Ga2O3 SBD performance. The research results reveal the impact of high-energy electron irradiation on the $\beta $ -Ga2O3 SBD, providing a basis for its application in harsh radiation environments.

The state of charge (SOC) trajectory planning for the power battery is the basis of realizing the global online optimal control of the plug-in hybrid-electric bus (PHEB) energy management system. This article proposed an adaptive rolling planning method for SOC trajectory of PHEBs power battery. First, a mathematical model for simulation research is established. Second, the driving cycles of PHEBs are collected, and the optimal SOC trajectories of driving cycles are obtained using dynamic programming (DP). In order to provide training data for the planning model of the optimal SOC trajectory of a trip segment (TS), the driving cycle and the optimal SOC trajectory of the driving cycle are cut into segments with a time length of 30 s. The incremental learning is used to construct a planning model of the TS’s optimal SOC trajectory and planning the optimal SOC trajectory of the actual TS. The optimal SOC trajectory of the actual TS is applied to proportion–integration-differentiation (PID) control and compared with PID based on mileage-based SOC planning (BMP). The result shows that this study can reduce fuel consumption by 0.8 L/100 km for the same driving cycle. Through incremental learning, the fuel consumption can be further reduced by 0.13 L/100 km.

The detection of bubbles in images of transmitter and receiver printed circuit boards (TR-PCBs) is a difficult task due to the need for precise subregional detection, fast processing speed, and high accuracy. The uneven illumination of the images further complicates the process of bubble segmentation. To address these issues, a TR-PCB bubble segmentation framework, the efficient channel and spatial attention (ECSA)-efficient feature-fused autoencoder (EFA)-dynamic threshold (DT) framework, is proposed, which has two stages. In the first stage, an ECSA module is designed to analyze complex background features by employing the complementary advantages of the channel and spatial position features. Then, an EFA is proposed for region segmentation in TR-PCB images. In the second stage, a DT approach based on maximizing the interclass variance and a sliding window is proposed to eliminate severe shadows and issues with uneven illumination in TR-PCB images. The DT approach segments the bubbles in each region and calculates the bubble rate more accurately than previous methods. The experimental results show that the proposed framework achieves superior performance in several quantitative metrics and is suitable for industrial applications, obtaining an average Dice score of 87.5% and an intersection over union (IoU) score of 77.8% on the region segmentation test set, and a 0.312 misclassification error (ME) value on the bubble segmentation test set.

The life prediction of proton exchange membrane fuel cell (PEMFC) is one of the methods to reduce the cost of using fuel cells. This helps to solve the problem of high using costs for fuel cell vehicles, which makes it difficult to commercialize them on a large scale. However, there is still a lack of real-time life prediction methods that can be applied to fuel cell vehicles. In this article, the first step is conducting actual road vehicle life experiments, which include urban roads, suburban roads, and highways. Collect experimental data and preprocess it to obtain a dataset of actual vehicle life. The second step is to divide the output power of the fuel cell into multiple operating conditions, calculate the voltage decay rate of each operating condition, and establish a real-time prediction (RTP) model for the fuel cell life. The third step is conducting fuel cell life prediction. The results show that compared with the Pei method, the RMSE and MAPE of the proposed RTP method for fuel cell life have decreased by 21% and 14% in the long-term prediction of 100 h, and the average reduction is 41% and 35% in the short-term prediction of 10 h.

The degradation of inverted metamorphic four-junction (GaInP/GaAs/In0.3Ga0.7As/In0.58Ga0.42As, IMM4J) solar cells irradiated by 1-MeV electrons was investigated via their spectral responses and the characterization of their electrical properties. As in the case of traditional three-junction (TJ) GaInP/GaAs/Ge solar cells, the electrical properties (Isc, Voc, and Pmax) decrease with the logarithmic change of the electron fluence. The degradation of open-circuit voltage (Voc) is slightly more pronounced than that of Isc in IMM4J solar cells because of the sum rule for Voc and the limit rule for Isc. The spectral response analysis showed that an In0.3Ga0.7As subcell was the most damaged subcell in the irradiated IMM4J solar cell, but an In0.58Ga0.42As subcell had the lowest initial short-circuit current density (Jsc), which was maintained throughout the irradiation test. We therefore focused on the In0.58Ga0.42As subcell. Deep-level transient spectroscopy (DLTS) experiments were realized to study emission and capture processes in two special full configurations of In0.58Ga0.42As and In0.3Ga0.7As subcells of the IMM4J solar cell. DLTS measurements reveal a dominant electron trap at 0.52 eV below the conduction band (Ec) of In0.58Ga0.42As, and the electron trap gradually evolved into Ec-0.46eV and Ec-0.58eV after 1-MeV electron irradiation. Based on the first-principles calculation, Ec-0.46 eV and Ec-0.58 eV can be assigned as ${\bf V}_{{\bf Ga}}^{\bf 0}/{\bf V}_{{\bf Ga}}^{{\rm{ - }}1}$ and ${\bf V}_{{\bf In}}^{\bf 0}/{\bf V}_{{\bf In}}^{{\bf - 1}}$, respectively. However, only one shallow level Ec-0.03eV was observed within the bandgap of In0.3Ga0.7As after irradiation with DLTS detection. We summarize the issues faced for the space application of IMM4J solar cells by comparing the spectral responses of IMM3J, IMM4J, and TJ solar cells. Finally, the optimization of the design and fabrication of IMM solar cells are proposed.

A novel strategy for the synthesis of 2-aroylthienothiazoles via C–H/N–O bond functionalization of ketoximes is developed. This reaction features excellent step- and atom-economy, as well as broad substrate scope. Various common ketoximes, even vinyl ketoximes, were efficiently converted to 2-aroylthienothiazoles. Preliminary mechanistic studies indicated that the radical process should be involved in this transformation. Moreover, the product exhibited good coordination with Cu(II), showing the potential application in the metal coordination field.

Background: With advances in science and technology, the application of artificial intelligence in medicine has significantly progressed. The purpose of this study is to explore whether the k-nearest neighbors (KNN) machine learning method can identify three milling states based on vibration signals: cancellous bone (CCB), ventral cortical bone (VCB), and penetration (PT) in robot-assisted cervical laminectomy. Methods: Cervical laminectomies were performed on the cervical segments of eight pigs using a robot. First, the bilateral dorsal cortical bone and part of the CCB were milled with a 5 mm blade and then the bilateral laminae were milled to penetration with a 2 mm blade. During the milling process using the 2 mm blade, the vibration signals were collected by the acceleration sensor, and the harmonic components were extracted using fast Fourier transform. The feature vectors were constructed with vibration signal amplitudes of 0.5, 1.0, and 1.5 kHz and the KNN was then trained by the features vector to predict the milling states. Results: The amplitudes of the vibration signals between VCB and PT were statistically different at 0.5, 1.0, and 1.5 kHz (P< 0.05), and the amplitudes of the vibration signals between CCB and VCB were significantly different at 0.5 and 1.5 kHz (P< 0.05). The KNN recognition success rates for the CCB, VCB, and PT were 92%, 98%, and 100%, respectively. A total of 6% and 2% of the CCB cases were identified as VCB and PT, respectively; 2% of VCB cases were identified as PT. Conclusions: The KNN can distinguish different milling states of a high-speed bur in robot-assisted cervical laminectomy based on vibration signals. This method is feasible for improving the safety of posterior cervical decompression surgery.

Human being perceives multiple tactile modalities in the process of sensation on the skin and interpretation in the brain. To date, several sensing techniques facilitate the accurate measurement of individual tactile modality, but multimodal static and dynamic sensing remain challenging. Moreover, low-cost and highly efficient interpretation techniques are still required for tactile perception. Herein, we present cost-effective and high-performing self-powered smart skins that mimic multimodal tactile perception, enabling accurate perception of pressure, vibration, and humidity in the process of sensation on the smart skin and interpretation by machine learning. The dynamic and static stimuli are encoded by triboelectric and hygroelectric principles in the smart skins, respectively, while the hygroscopic nature empowers humidity sensation capability in the smart skin with an accuracy rate as high as 84.0%−100.0%. We believe our smart skin will enable the smooth transition of e-skin into practical applications, such as robotics, prosthetics, healthcare, and intelligent industry.

Herein, 1‐MeV electron irradiation is conducted to the AlGaInP (1.89 eV)/AlInGaAs (1.45 eV)/InGaAs (1.14 eV)/Ge (0.67 eV) upright metamorphic four‐junction (UMM4J) solar cell and its upper three key subcells. Light IV (LIV), external quantum efficiency (EQE), and deep level transient spectroscopy (DLTS) measurements are implemented to analyze the degradation of electric properties and evolution of defects of the UMM4J solar cell. The LIV results illustrate that the short circuit current (Isc), open circuit voltage (Voc), and maximum power (Pmax) degrade as a function of the logarithmic change of the electron fluence. EQE curves confirm the excellent anti‐irradiation characteristic of the AlGaInP (1.89 eV) subcell. Three types of electron traps E1(Ec–0.16 eV), E2(Ec–0.47 eV), and E3(Ec–0.55 eV) are detected in the AlGaInP subcell. Two types of hole traps H1(Ev+ 0.35 eV) and H2(Ev+ 0.70 eV) are discovered in the AlInGaAs subcell and two types of hole traps H3(Ev+ 0.05 eV) and H4(Ev+ 0.36 eV) are observed in the InGaAs subcell. Nonionizing energy loss (NIEL) of 1‐MeV electrons in are calculated via Screened Relativist method. The defect concentrations show an approximately linear relationship with the displacement dose. InGaAs (1.14 eV) subcell is the current limiting junction of UMM4J solar cells for space applications. In InGaAs diodes, a characteristic peak of shallow level majority carrier defect (H3) is observed at Ev+ 0.05 eV, and the defect concentration increases significantly with the electron fluence. Furthermore, the initial majority carrier defect (H4, Ev+ 0.36 eV) increases slightly.

Low surface energy materials resist adhesion due to their chemical inertness and non-wetting properties. Herein, we report the creation of a transparent ionogel adhesive that uses ion–dipole interactions to achieve a higher bonding performance to polytetrafluoroethylene (PTFE) relative to most commercial glues. The ionogel adhesive is composed of a poly(hexafluorobutyl acrylate-co-methyl methacrylate) random copolymer and a hydrophobic ionic liquid. The prepared ionogel can adhere to various hydrophobic substrates, such as PTFE, polypropylene, and polyethylene, as well as hydrophilic glass, ceramics, and steel. The design strategy and adhesion behavior are well interpreted using the density functional theory calculations and molecular dynamics simulations. The straightforward ultraviolet-curing method, high optical clarity, versatile adhesion ability, and reversible adhesion capabilities make this high-performance adhesive a promising product for commercialization.

PD-1 blockade is highly effective in patients with mismatch repair-deficient or microsatellite instability-high metastatic colorectal cancer. The role of single-agent PD-1 blockade in the neoadjuvant setting for resectable mismatch repair-deficient or microsatellite instability-high colorectal cancer remains unclear. We investigated the efficacy and safety of PD-1 blockade with toripalimab, with or without the COX-2 inhibitor celecoxib, as neoadjuvant treatment for mismatch repair-deficient or microsatellite instability-high, locally advanced, colorectal cancers.

As the low-cost, natural multi-component for elemental doping and environment-friendly characteristics, biomass-derived porous carbon for energy storage attracts intense attention. Herein, walnut shells-based porous carbon has been obtained through carbonization, hydrothermal and activation treatment. The corresponding porous carbon owns superior electrochemical performances with specific capacitance reaching up to 462Fg−1at 1Ag−1, and shows excellent cycling stability (5000 cycles, ∼94.2% of capacitance retention at 10Ag−1). Moreover, the symmetry supercapacitor achieves high specific capacitance (197Fg−1at 1Ag−1), relevant electrochemical cycling stability (5000 cycles, 89.2% of capacitance retention at 5Ag−1) and high power/energy density (42.8Whkg−1at 1249Wkg−1). Therefore, the facile synthesis approach and superb electrochemical performance ensure that the walnut shells-derived porous carbon is a promising electrode material candidate for supercapacitors.