Semantic segmentation is a key step in scene understanding for autonomous driving. Although deep learning has significantly improved the segmentation accuracy, current high-quality models such as ...PSPNet and DeepLabV3 are inefficient given their complex architectures and reliance on multi-scale inputs. Thus, it is difficult to apply them to real-time or practical applications. On the other hand, existing real-time methods cannot yet produce satisfactory results on small objects such as traffic lights, which are imperative to safe autonomous driving. In this paper, we improve the performance of real-time semantic segmentation from two perspectives, methodology and data. Specifically, we propose a real-time segmentation model coined Narrow Deep Network (NDNet) and build a synthetic dataset by inserting additional small objects into the training images. The proposed method achieves 65.7% mean intersection over union (mIoU) on the Cityscapes test set with only 8.4G floating-point operations (FLOPs) on <inline-formula> <tex-math notation="LaTeX">1024\times 2048 </tex-math></inline-formula> inputs. Furthermore, by re-training the existing PSPNet and DeepLabV3 models on our synthetic dataset, we obtained an average 2% mIoU improvement on small objects.
Amyloid-β (Aβ) accumulation in the brain is a pivotal event in the pathogenesis of Alzheimer's disease (AD), and its clearance from the brain is impaired in sporadic AD. Previous studies suggest that ...approximately half of the Aβ produced in the brain is cleared by transport into the periphery. However, the mechanism and pathophysiological significance of peripheral Aβ clearance remain largely unknown. The kidney is thought to be responsible for Aβ clearance, but direct evidence is lacking. In this study, we investigated the impact of unilateral nephrectomy on the dynamic changes in Aβ in the blood and brain in both humans and animals and on behavioural deficits and AD pathologies in animals. Furthermore, the therapeutic effects of the diuretic furosemide on Aβ clearance via the kidney were assessed. We detected Aβ in the kidneys and urine of both humans and animals and found that the Aβ level in the blood of the renal artery was higher than that in the blood of the renal vein. Unilateral nephrectomy increased brain Aβ deposition; aggravated AD pathologies, including Tau hyperphosphorylation, glial activation, neuroinflammation, and neuronal loss; and aggravated cognitive deficits in APP/PS1 mice. In addition, chronic furosemide treatment reduced blood and brain Aβ levels and attenuated AD pathologies and cognitive deficits in APP/PS1 mice. Our findings demonstrate that the kidney physiologically clears Aβ from the blood, suggesting that facilitation of Aβ clearance via the kidney represents a novel potential therapeutic approach for AD.
Although deep learning has achieved great success in many computer vision tasks, its performance relies on the availability of large datasets with densely annotated samples. Such datasets are ...difficult and expensive to obtain. In this article, we focus on the problem of learning representation from unlabeled data for semantic segmentation. Inspired by two patch-based methods, we develop a novel self-supervised learning framework by formulating the jigsaw puzzle problem as a patch-wise classification problem and solving it with a fully convolutional network. By learning to solve a jigsaw puzzle comprising 25 patches and transferring the learned features to semantic segmentation task, we achieve a 5.8% point improvement on the Cityscapes dataset over the baseline model initialized from random values. It is noted that we use only about 1/6 training images of Cityscapes in our experiment, which is designed to imitate the real cases where fully annotated images are usually limited to a small number. We also show that our self-supervised learning method can be applied to different datasets and models. In particular, we achieved competitive performance with the state-of-the-art methods on the PASCAL VOC2012 dataset using significantly fewer time costs on pretraining.
A large circular transmitter (power pad) is often utilized for inductively wireless charging so that the power receiver can position itself freely within the power pad. However, the uneven magnetic ...field distribution leads to the impedance mismatch and the variable transferred power due to the misalignment between a receiving coil and a relatively large transmitter coil. In this paper, the effects of the turn numbers, trace spacing, and coil size on the uniformity of magnetic field distribution over a planar spiral coil (PSC) are investigated based on an analytical model. Then, the genetic algorithm is used to optimize the trace width and spacing of a ten-turn PSC with the fixed outer and inner diameters, whereas the magnetic field is numerically calculated by HFSS software. The geometrically optimal PSC is obtained with the trace width approximately equal to trace spacing for each turn by minimizing the coefficient of variation (COV) of magnetic field within an effective charging area over the coil. The results showed that the simulated and measured COVs of magnetic field were 0.130 and 0.121, respectively, which are obviously less than that of the uniformly spaced regular coil and show a little less than that of the optimally spaced coil with fixed trace width. Therefore, the trace spacing is a major consideration to achieve the uniform magnetic field, whereas the trace width variation is applied to refine the evenness of field. It is also expected that the proposed GA-based optimization is well applied to design other Txs with conformable shape in inductively WPT systems for stable power delivery regardless of the receiver positions.
Bacterial wilt, caused by
, one of the most destructive phytopathogens, leads to significant annual crop yield losses. Type III effectors (T3Es) mainly contribute to the virulence of
, usually by ...targeting immune-related proteins. Here, we clarified the effect of a novel E3 ubiquitin ligase (NEL) T3E, RipAW, from
on pathogen-associated molecular pattern (PAMP)-triggered immunity (PTI) and further explored its action mechanism. In the susceptible host
, we monitored the expression of PTI marker genes, flg22-induced ROS burst, and callose deposition in
- and
-transgenic plants. Our results demonstrated that RipAW suppressed host PTI in an NEL-dependent manner. By Split-Luciferase Complementation, Bimolecular Fluorescent Complimentary, and Co-Immunoprecipitation assays, we further showed that RipAW associated with three crucial components of the immune receptor complex, namely FLS2, XLG2, and BIK1. Furthermore, RipAW elevated the ubiquitination levels of FLS2, XLG2, and BIK1, accelerating their degradation via the 26S proteasome pathway. Additionally, co-expression of FLS2, XLG2, or BIK1 with RipAW partially but significantly restored the RipAW-suppressed ROS burst, confirming the involvement of the immune receptor complex in RipAW-regulated PTI. Overall, our results indicate that RipAW impairs host PTI by disrupting the immune receptor complex. Our findings provide new insights into the virulence mechanism of
.
To develop an artificial intelligence (AI)-based algorithm which can automatically detect food items from images acquired by an egocentric wearable camera for dietary assessment.
To study human diet ...and lifestyle, large sets of egocentric images were acquired using a wearable device, called eButton, from free-living individuals. Three thousand nine hundred images containing real-world activities, which formed eButton data set 1, were manually selected from thirty subjects. eButton data set 2 contained 29 515 images acquired from a research participant in a week-long unrestricted recording. They included both food- and non-food-related real-life activities, such as dining at both home and restaurants, cooking, shopping, gardening, housekeeping chores, taking classes, gym exercise, etc. All images in these data sets were classified as food/non-food images based on their tags generated by a convolutional neural network.
A cross data-set test was conducted on eButton data set 1. The overall accuracy of food detection was 91·5 and 86·4 %, respectively, when one-half of data set 1 was used for training and the other half for testing. For eButton data set 2, 74·0 % sensitivity and 87·0 % specificity were obtained if both 'food' and 'drink' were considered as food images. Alternatively, if only 'food' items were considered, the sensitivity and specificity reached 85·0 and 85·8 %, respectively.
The AI technology can automatically detect foods from low-quality, wearable camera-acquired real-world egocentric images with reasonable accuracy, reducing both the burden of data processing and privacy concerns.
A relationship between triboelectric charge and contact force for two triboelectric layers is presented, by combining the theories of insulator contact charging and contact mechanics. Experimental ...verification has been successfully performed using contact-mode triboelectric nanogenerators (TENGs) in two cases: (a) under varying contact forces while keeping the surface roughness profile constant, and (b) under varying surface roughness profiles while keeping the contact force constant. The theory presented here can serve as an important guide in the design of triboelectric systems, particularly of a contact-mode TENG structure for specific applications and self-powered systems.
•A triboelectric charge (q) - contact force (F) relationship is developed and verified using triboelectric nanogenerators.•Experimental verification is successfully performed under varying contact forces and surface roughness profiles.•The output voltages of triboelectric nanogenerators are predicted within one order of magnitude in both experimental cases.•The presented q−F relationship can serve as an important guide in the design of triboelectric systems.
The rapid development of autonomous driving in recent years presents many challenges for scene understanding. As an essential step towards scene understanding, semantic segmentation has received ...increased attention in the past few years. Although deep learning based approaches have achieved great success in improving the segmentation accuracy, most of them suffer from an inefficiency problem and can hardly be applied to real-time applications. In this paper, we analyze the computational cost of Convolutional Neural Network (CNN) and find that the inefficiency of CNNs is mainly caused by their wide structure rather than deep structure. In addition, the success of pruning based model compression methods proves that there are many redundant channels in CNNs. Thus, we design a narrow while deep backbone network to improve the efficiency of semantic segmentation. By casting our network to fully convolutional network (FCN32) segmentation architecture, the basic structure of most segmentation methods, we achieve 61.5% mIoU on Cityscapes validation dataset with only 4.2G floating-point operations (FLOPs) on <inline-formula> <tex-math notation="LaTeX">1024\times 2048 </tex-math></inline-formula> inputs, which already outperforms one of the earliest real-time deep learning based segmentation methods: ENet (58.3% mIoU, 3.8G FLOPs on <inline-formula> <tex-math notation="LaTeX">640\times 360 </tex-math></inline-formula> inputs). By further refining the output resolution of our network to the 1/8 of the input resolution with a simple encoder-decoder structure, we achieve 65.3% mIoU on Cityscapes test set with 14.0G FLOPs and 39.9 frames per second (FPS) on Titan X card. We have made our model publicly available at https://github.com/zgyang-hnu/NDNet .
How glasses relax at room temperature is still a great challenge for both experimental and simulation studies due to the extremely long relaxation time-scale. Here, by employing a modified molecular ...dynamics simulation technique, we extend the quantitative measurement of relaxation process of metallic glasses to room temperature. Both energy relaxation and dynamics, at low temperatures, follow a stretched exponential decay with a characteristic stretching exponent β = 3/7, which is distinct from that of supercooled liquid. Such aging dynamics originates from the release of energy, an intrinsic nature of out-of-equilibrium system, and manifests itself as the elimination of defects through localized atomic strains. This finding is also supported by long-time stress-relaxation experiments of various metallic glasses, confirming its validity and universality. Here, we show that the distinct relaxation mechanism can be regarded as a direct indicator of glass transition from a dynamic perspective.