Humans can naturally and effectively find salient regions in complex scenes. Motivated by this observation, attention mechanisms were introduced into computer vision with the aim of imitating this ...aspect of the human visual system. Such an attention mechanism can be regarded as a dynamic weight adjustment process based on features of the input image. Attention mechanisms have achieved great success in many visual tasks, including image classification, object detection, semantic segmentation, video understanding, image generation, 3D vision, multimodal tasks, and self-supervised learning. In this survey, we provide a comprehensive review of various attention mechanisms in computer vision and categorize them according to approach, such as channel attention, spatial attention, temporal attention, and branch attention; a related repository
https://github.com/MenghaoGuo/Awesome-Vision-Attentions
is dedicated to collecting related work. We also suggest future directions for attention mechanism research.
Wireless sensor networks (WSNs) are an important tool for monitoring distributed remote environments. As one of the key technologies involved in WSNs, node fault detection is indispensable in most ...WSN applications. It is well known that the distributed fault detection (DFD) scheme checks out the failed nodes by exchanging data and mutually testing among neighbor nodes in this network., but the fault detection accuracy of a DFD scheme would decrease rapidly when the number of neighbor nodes to be diagnosed is small and the node's failure ratio is high. In this paper, an improved DFD scheme is proposed by defining new detection criteria. Simulation results demonstrate that the improved DFD scheme performs well in the above situation and can increase the fault detection accuracy greatly.
The class activation maps are generated from the final convolutional layer of CNN. They can highlight discriminative object regions for the class of interest. These discovered object regions have ...been widely used for weakly-supervised tasks. However, due to the small spatial resolution of the final convolutional layer, such class activation maps often locate coarse regions of the target objects, limiting the performance of weakly-supervised tasks that need pixel-accurate object locations. Thus, we aim to generate more fine-grained object localization information from the class activation maps to locate the target objects more accurately. In this paper, by rethinking the relationships between the feature maps and their corresponding gradients, we propose a simple yet effective method, called LayerCAM. It can produce reliable class activation maps for different layers of CNN. This property enables us to collect object localization information from coarse (rough spatial localization) to fine (precise fine-grained details) levels. We further integrate them into a high-quality class activation map, where the object-related pixels can be better highlighted. To evaluate the quality of the class activation maps produced by LayerCAM, we apply them to weakly-supervised object localization and semantic segmentation. Experiments demonstrate that the class activation maps generated by our method are more effective and reliable than those by the existing attention methods. The code will be made publicly available.
High‐performance uncooled photodetectors operating in the long‐wavelength infrared and terahertz regimes are highly demanded in the military and civilian fields. Photothermoelectric (PTE) detectors, ...which combine photothermal and thermoelectric conversion processes, can realize ultra‐broadband photodetection without the requirement of a cooling unit and external bias. In the last few decades, the responsivity and speed of PTE‐based photodetectors have made impressive progress with the discovery of novel thermoelectric materials and the development of nanophotonics. In particular, by introducing hot‐carrier transport into low‐dimensional material–based PTE detectors, the response time has been successfully pushed down to the picosecond level. Furthermore, with the assistance of surface plasmon, antenna, and phonon absorption, the responsivity of PTE detectors can be significantly enhanced. Beyond the photodetection, PTE effect can also be utilized to probe exotic physical phenomena in spintronics and valleytronics. Herein, recent advances in PTE detectors are summarized, and some potential strategies to further improve the performance are proposed.
The room‐temperature detection of long‐wavelength infrared and terahertz radiation can be realized by photothermoelectric (PTE) detectors. The responsivity and the response speed of PTE‐based photodetectors have made impressive progress with the discovery of novel thermoelectric materials and the development of nanophotonics. Beyond light detection, the PTE effect can be utilized to study novel physical phenomena in spintronics and valleytronics.
Abstract
The self-powered and ultra-broadband photodetectors based on photothermoelectric (PTE) effect are promising for diverse applications such as sensing, environmental monitoring, night vision ...and astronomy. The sensitivity of PTE photodetectors is determined by the Seebeck coefficient and the rising temperature under illumination. Previous PTE photodetectors mostly rely on traditional thermoelectric materials with Seebeck coefficients in the range of 100 μV K
−1
, and array structures with multiple units are usually employed to enhance the photodetection performance. Herein, we demonstrate a reduced SrTiO
3
(r-STO) based PTE photodetector with sensitivity up to 1.2 V W
−1
and broadband spectral response from 325 nm to 10.67 μm. The high performance of r-STO PTE photodetector is attributed to its intrinsic high Seebeck coefficient and phonon-enhanced photoresponse in the long wavelength infrared region. Our results open up a new avenue towards searching for novel PTE materials beyond traditional thermoelectric materials for low-cost and high-performance photodetector at room temperature.
•Impacts of COVID-19 on energy demand and consumption have been substantial.•The changes in energy intensity (GDP/Mtoe) presented spatial-temporal differences.•System thinking is recommended to ...analyse how to stabilise energy demand.•The energy recovery presents heterogeneous characteristics in countries/regions.•The rebound effects of digitalisation in energy consumption need to be assessed.
COVID-19 has caused great challenges to the energy industry. Potential new practices and social forms being facilitated by the pandemics are having impacts on energy demand and consumption. Spatial and temporal heterogeneities of impacts appear gradually due to the dynamics of pandemics and mitigation measures. This paper overviews the impacts and challenges of COVID-19 pandemics on energy demand and consumption and highlights energy-related lessons and emerging opportunities. The discussion on energy-related issues is divided into four main sections: emergency situation and its impacts, environmental impacts and stabilising energy demand, recovering energy demand, and lessons and emerging opportunities. The changes in energy requirements are compared and analysed from multiple perspectives according to available data and information. In general, although the overall energy demand declines, the spatial and temporal variations are complicated. The energy intensity has presented apparent changes, the extra energy for COVID-19 fighting is non-negligible for stabilising energy demand, and the energy recovery in different regions presents significant differences. A crucial issue has been to allocate and find energy-related emerging opportunities for the post pandemics. This study could offer a direction in opening new avenues for increasing energy efficiency and promoting energy saving.
Delving Deep Into Label Smoothing Zhang, Chang-Bin; Jiang, Peng-Tao; Hou, Qibin ...
IEEE transactions on image processing,
2021, Letnik:
30
Journal Article
Recenzirano
Odprti dostop
Label smoothing is an effective regularization tool for deep neural networks (DNNs), which generates soft labels by applying a weighted average between the uniform distribution and the hard label. It ...is often used to reduce the overfitting problem of training DNNs and further improve classification performance. In this paper, we aim to investigate how to generate more reliable soft labels. We present an Online Label Smoothing (OLS) strategy, which generates soft labels based on the statistics of the model prediction for the target category. The proposed OLS constructs a more reasonable probability distribution between the target categories and non-target categories to supervise DNNs. Experiments demonstrate that based on the same classification models, the proposed approach can effectively improve the classification performance on CIFAR-100, ImageNet, and fine-grained datasets. Additionally, the proposed method can significantly improve the robustness of DNN models to noisy labels compared to current label smoothing approaches. The source code is available at our project page: https://mmcheng.net/ols/
Energy metabolism is significantly reprogrammed in many human cancers, and these alterations confer many advantages to cancer cells, including the promotion of biosynthesis, ATP generation, ...detoxification and support of rapid proliferation. The pentose phosphate pathway (PPP) is a major pathway for glucose catabolism. The PPP directs glucose flux to its oxidative branch and produces a reduced form of nicotinamide adenine dinucleotide phosphate (NADPH), an essential reductant in anabolic processes. It has become clear that the PPP plays a critical role in regulating cancer cell growth by supplying cells with not only ribose-5-phosphate but also NADPH for detoxification of intracellular reactive oxygen species, reductive biosynthesis and ribose biogenesis. Thus, alteration of the PPP contributes directly to cell proliferation, survival and senescence. Furthermore, recent studies have shown that the PPP is regulated oncogenically and/or metabolically by numerous factors, including tumor suppressors, oncoproteins and intracellular metabolites. Dysregulation of PPP flux dramatically impacts cancer growth and survival. Therefore, a better understanding of how the PPP is reprogrammed and the mechanism underlying the balance between glycolysis and PPP flux in cancer will be valuable in developing therapeutic strategies targeting this pathway.
The scalable production of hydrogen could conveniently be realized by alkaline water electrolysis. Currently, the major challenge confronting hydrogen evolution reaction (HER) is lacking inexpensive ...alternatives to platinum-based electrocatalysts. Here we report a high-efficient and stable electrocatalyst composed of ruthenium and cobalt bimetallic nanoalloy encapsulated in nitrogen-doped graphene layers. The catalysts display remarkable performance with low overpotentials of only 28 and 218 mV at 10 and 100 mA cm
, respectively, and excellent stability of 10,000 cycles. Ruthenium is the cheapest platinum-group metal and its amount in the catalyst is only 3.58 wt.%, showing the catalyst high activity at a very competitive price. Density functional theory calculations reveal that the introduction of ruthenium atoms into cobalt core can improve the efficiency of electron transfer from alloy core to graphene shell, beneficial for enhancing carbon-hydrogen bond, thereby lowing ΔG
of HER.
Alternaria leaf spot, Brown spot, Mosaic, Grey spot, and Rust are five common types of apple leaf diseases that severely affect apple yield. However, the existing research lacks an accurate and fast ...detector of apple diseases for ensuring the healthy development of the apple industry. This paper proposes a deep learning approach that is based on improved convolutional neural networks (CNNs) for the real-time detection of apple leaf diseases. In this paper, the apple leaf disease dataset (ALDD), which is composed of laboratory images and complex images under real field conditions, is first constructed via data augmentation and image annotation technologies. Based on this, a new apple leaf disease detection model that uses deep-CNNs is proposed by introducing the GoogLeNet Inception structure and Rainbow concatenation. Finally, under the hold-out testing dataset, using a dataset of 26,377 images of diseased apple leaves, the proposed INAR-SSD (SSD with Inception module and Rainbow concatenation) model is trained to detect these five common apple leaf diseases. The experimental results show that the INAR-SSD model realizes a detection performance of 78.80% mAP on ALDD, with a high-detection speed of 23.13 FPS. The results demonstrate that the novel INAR-SSD model provides a high-performance solution for the early diagnosis of apple leaf diseases that can perform real-time detection of these diseases with higher accuracy and faster detection speed than previous methods.