Skeleton-based human action recognition is becoming popular due to its computational efficiency and robustness. Since not all skeleton joints are informative for action recognition, attention ...mechanisms are adopted to extract informative joints and suppress the influence of irrelevant ones. However, existing attention frameworks usually ignore helpful scenario context information. In this paper, we propose a cross-attention module that consists of a self-attention branch and a cross-attention branch for skeleton-based action recognition. It helps to extract joints that are not only more informative but also highly correlated to the corresponding scenario context information. Moreover, the cross-attention module maintains input variables' size and can be flexibly incorporated into many existing frameworks without breaking their behaviors. To facilitate end-to-end training, we further develop a scenario context information extraction branch to extract context information from raw RGB video directly. We conduct comprehensive experiments on the NTU RGB+D and the Kinetics databases, and experimental results demonstrate the correctness and effectiveness of the proposed model.
A number of advanced image editing technologies have demonstrated impressive performance in synthesizing visually pleasing results in accordance with user instructions. In this paper, we further ...extend the practicalities of image editing technology by proposing the conditional image repainting (CIR) task, which requires the model to synthesize realistic visual content based on multiple cross-modality conditions provided by the user. We first define condition inputs and formulate two-phased CIR models as the baseline. After that, we further design unified CIR models with novel condition fusion modules to improve the performance. For allowing users to express their intent more freely, our CIR models support both attributes and language to represent colors of repainted visual content. We demonstrate the effectiveness of CIR models by collecting and processing four datasets. Finally, we present a number of practical application scenarios of CIR models to demonstrate its usability.
Perovskite quantum dots (PQDs), as the promising materials for the blue light-emitting diodes (LEDs), own huge advantages as ultra-high color purity, flexibility and whole-spectrum tunability. ...Through dimensional and compositional engineering, PQD-LEDs have shown superiority in deep-blue light emission. However, compared with the fast development of red and green PeLEDs, the electroluminescent performance of PQD-LEDs has faced more obstacles. In this review, we aim to explore and state the uniqueness and the possible solutions for the bottleneck problems of the PQD-LEDs.
In this paper, we explore synthesizing person images with multiple conditions for various backgrounds. To this end, we propose a framework named ``MISC" for conditional image generation and image ...compositing. For conditional image generation, we improve the existing condition injection mechanisms by leveraging the inter-condition correlations. For the image compositing, we theoretically prove the weaknesses of the cutting-edge methods, and make it more robust by removing the spatially-invariance constraint, and enabling the bounding mechanism and the spatial adaptability. We show the effectiveness of our method on the Video Instance-level Parsing dataset, and demonstrate the robustness through controllability tests.
Two-dimensional (2D) Dion-Jacobson (DJ) halide perovskites are being intensively investigated owing to their superior stability compared with 2D Ruddlesden-Popper (RP) perovskites. Although various ...organic cations have been reported to form 2D DJ perovskites, their structure-property relationships are rarely studied. In this work, two novel organic cations MDAN (4,4′-methylenedianilinium) and EDAN (4,4′-ethylenedianilinium) were applied to form quasi-2D DJ perovskites with nominal layer number
n
= 6. An (EDAN)MA
5
Pb
6
I
19
:0.15MACl based inverted solar cell achieved a power conversion efficiency (PCE) of 13.2%, outperforms the 9.6% PCE achieved by a (MDAN)MA
5
Pb
6
I
19
:0.15MACl based device. The formation of quasi-2D perovskites with high-
n
values was observed for both compounds, while the superior performance is demonstrated by the (EDAN)MA
5
Pb
6
I
19
:0.15MACl based device, which could be attributed to its better film morphology and lower defect density. Intriguingly, we found that EDAN based quasi-2D perovskites were evenly distributed in the film while MDAN based quasi-2D perovskites concentrated near the film surface, accounting for an unfavorable electronic band alignment of the latter. Both (MDAN/EDAN)MA
5
Pb
6
I
19
:0.15MACl films remained unchanged in air with 50% relative humidity for 20 days. This work not only presents a decent device performance, but also explores understanding of tuning morphology towards high-efficiency quasi-2D Dion-Jacobson perovskite solar cells.
New small molecules MDAN and EDAN are synthesized and found to be different in tuning the morphology of low-dimensional perovskite morphology.
Conditional image repainting (CIR) is an advanced image editing task, which requires the model to generate visual content in user-specified regions conditioned on multiple cross-modality constraints, ...and composite the visual content with the provided background seamlessly. Existing methods based on two-phase architecture design assume dependency between phases and cause color-image incongruity. To solve these problems, we propose a novel Unified Conditional image Repainting Network (UniCoRN). We break the two-phase assumption in the CIR task by constructing the interaction and dependency relationship between background and other conditions. We further introduce the hierarchical structure into cross-modality similarity model to capture feature patterns at different levels and bridge the gap between visual content and color condition. A new Landscape-CIR dataset is collected and annotated to expand the application scenarios of the CIR task. Experiments show that UniCoRN achieves higher synthetic quality, better condition consistency, and more realistic compositing effect.
Perovskite solar cells (PSCs) are widely studied as the most promising photovoltaic technology, but their performance is sensitive to the morphology and the defect density of the perovskite films. ...Herein, additive engineering strategy is applied to further improve the film morphology and device performance by doping a small amount of natural product, D‐Aspartic acid (D‐2‐Aminobutanedioic acid, D‐Asp), into perovskite precursor solution. The modified device exhibits a greatly enhanced power conversion efficiency of 22.7%, which is unprecedented for PSCs doped with natural amino acids, and 3000 h stability in ambient air is achieved. Through systematical characterizations, it is concluded that D‐Asp could result in thicker film formation, outstanding perovskite film morphology, and reduced defect sites passivated by the multiple functional groups. The results prove that the performance and stability of the state‐of‐the‐art mixed ion PSCs with regular architecture could be effectively enhanced by D‐Asp. The minimum usage of the natural product as an additive is beneficial for the fabrication process and cost‐control in the industrialization of PSCs. This work also highlights the different passivating mechanisms for the molecules with multiple functional groups which are meaningful for materials design.
A natural product, D‐Aspartic acid, is applied as an additive for perovskite solar cells which effectively improves film morphology and passivates the defects of the perovskite, thus greatly enhancing the power conversion efficiency (PCE) and stability of the device.
Language-based colorization produces plausible colors consistent with the language description provided by the user. Recent studies introduce additional annotation to prevent color-object coupling ...and mismatch issues, but they still have difficulty in distinguishing instances corresponding to the same object words. In this paper, we propose a transformer-based framework to automatically aggregate similar image patches and achieve instance awareness without any additional knowledge. By applying our presented luminance augmentation and counter-color loss to break down the statistical correlation between luminance and color words, our model is driven to synthesize colors with better descriptive consistency. We further collect a dataset to provide distinctive visual characteristics and detailed language descriptions for multiple instances in the same image. Extensive experiments demonstrate our advantages of synthesizing visually pleasing and description-consistent results of instance-aware colorization.
A vertical integration of indium gallium nitride/gallium nitride (InGaN/GaN)-based microlight-emitting diode (μLED) and GaN ultraviolet bipolar junction transistor (BJT) phototransistor (UVPT), based ...on the same GaN material and process platform (UVPT-μLED), was proposed. The integrated device is a novel integrated device with light-emitting, detecting, sensing, driving and regulating functions. It can be used as a receiver and transmitter. The light-emission effect of the μLED on UVPT can be regulated by changing the power of an external light. Optical-electrical-optical conversion was conducted with Silvaco TCAD software. Electro-optic modulation characteristics and optical power amplification were attained through a Silvaco TCAD simulation. By optimization of the size and concentration of the device epitaxial, the current gain of the optimized device reaches a maximum of 240. The UVPT-μLED device shows a strong response to small-power light and a stable response to high-power light. The UVPT-μLED chip with a high switching ratio was successfully prepared and used to verify the feasibility of the UVPT-μLED device. The proposed monolithic integrated device not only can be used for traditional displays but also shows great potential in intelligent displays, such as human–computer interactive systems, on-chip optical interconnects, photonic chips and visible-light communications.
Low-dimensional organic–inorganic metal halide hybrids exhibit promising optical properties for light emitting applications. However, developing lead-free blue-light emitters with high ...photoluminescence quantum efficiency (PLQE) remains an ongoing challenge. In our work, a novel zero-dimensional (0D) indium hybrid compound (MA) 4 InCl 7 (MA = CH 3 NH 3 + ) was developed which exhibited broadband blue emission with a PLQE of 11.2% when excited by ultraviolet (UV) light. More interestingly, upon Cs + or Mn 2+ doping, the emission of the 0D compound further blue-shifted and became narrower, while the PLQE was significantly enhanced to 18.8% (Cs + ) or 20.7% (Mn 2+ ). More prominent PLQE enhancement to 74.7% was observed after Sb 3+ doping, which also altered the emission spectrum to the orange region. According to experimental characterization and theoretical calculations, we attribute the PLQE enhancement upon Cs + and Mn 2+ doping to defect passivation and the orange emission upon Sb 3+ doping to altered emission centers. We have demonstrated that multiple metal ions possess the ability to improve the light emitting properties of 0D organic–inorganic metal halides, and (MA) 4 InCl 7 could be utilized as a Sb 3+ heavy metal ion sensor with high selectivity and sensitivity.
PDF
