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.
Aerosol optical depth (AOD) is widely recognized as a critical indicator in understanding atmospheric physics and regional air quality because of its capability for quantifying aerosol loading in the ...atmosphere. Retrieving AOD from space-borne sensors' observations has become the primary technique for monitoring aerosol loading on a large scale. There is currently a renewed interest in designing new satellite sensors and developing more advanced retrieval algorithms to measure AOD from space in order to better quantify concentrations of particulate matters (PMs) for advanced air quality management, environmental health assessment, and climate change studies. However, retrieving high-resolution AOD at varying scales is still a challenging task due to the low signal-to-noise ratio in sensing, algorithmic synthesis constraints, downscaling issues, and data gaps resulting from adverse impacts such as cloud contamination. Current state-of-the-art technologies still do not permit delicate urban-scale environmental health studies based on appropriate AOD-PMs relationships. This review paper provides a holistic view of the major advances in AOD measurements, elucidates the limitations of current AOD products, presents the challenges with respect to the derivation of high-resolution AOD, and highlights perspectives regarding the possible improvements of satellite-based AOD estimation.
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/
Deep Hough Transform for Semantic Line Detection Zhao, Kai; Han, Qi; Zhang, Chang-Bin ...
IEEE transactions on pattern analysis and machine intelligence,
09/2022, Letnik:
44, Številka:
9
Journal Article
Recenzirano
Odprti dostop
We focus on a fundamental task of detecting meaningful line structures, a.k.a. , semantic line, in natural scenes. Many previous methods regard this problem as a special case of object detection and ...adjust existing object detectors for semantic line detection. However, these methods neglect the inherent characteristics of lines, leading to sub-optimal performance. Lines enjoy much simpler geometric property than complex objects and thus can be compactly parameterized by a few arguments. To better exploit the property of lines, in this paper, we incorporate the classical Hough transform technique into deeply learned representations and propose a one-shot end-to-end learning framework for line detection. By parameterizing lines with slopes and biases, we perform Hough transform to translate deep representations into the parametric domain, in which we perform line detection. Specifically, we aggregate features along candidate lines on the feature map plane and then assign the aggregated features to corresponding locations in the parametric domain. Consequently, the problem of detecting semantic lines in the spatial domain is transformed into spotting individual points in the parametric domain, making the post-processing steps, i.e., non-maximal suppression, more efficient. Furthermore, our method makes it easy to extract contextual line features that are critical for accurate line detection. In addition to the proposed method, we design an evaluation metric to assess the quality of line detection and construct a large scale dataset for the line detection task. Experimental results on our proposed dataset and another public dataset demonstrate the advantages of our method over previous state-of-the-art alternatives. The dataset and source code is available at https://mmcheng.net/dhtline/ .
This paper presents a thorough review of control technologies that have been applied to wastewater treatment processes in the environmental engineering regime in the past four decades. It aims to ...provide a comprehensive technological review for both water engineering professionals and control specialists, giving rise to a suite of up-to-date pathways to impact this field in light of the classified technology hubs. The assessment was conducted with respect to linear control, linearizing control, nonlinear control, and artificial intelligence-based control. The application domain of each technology hub was summarized into a set of comparative tables for a holistic assessment. Challenges and perspectives were offered to these field engineers to help orient their future endeavor.
Densely deployed cellular wireless networks, which employ small cell technology, are being widely implemented. Mitigating the impact of inter- and intracell signal interferences induced by the ...operations of these networks is a challenging yet essential task. In this paper, we consider adaptive rate scheduling for a transmitting node, regardless of whether it is a base station (BS) or a mobile user. We aim to maximize the system's throughput through the employment of fractional frequency reuse (FFR) schemes. Each BS employs either an omnidirectional or a directional antenna system. We derive the optimal configuration of the FFR scheme and evaluate the ensuing system's performance behavior under absolute and proportional fairness requirements. To maximize the attained throughput by mobiles, we determine the best method to use to classify cell users into interior and edge mobiles. The bandwidth levels allocated for serving interior and edge mobiles are optimized. We derive approximate closed-form mathematical expressions for calculating the probability distributions of the interference signal levels measured at the destined receivers. We account for the impact of the classification process on intercell interference power levels. Under an absolute fairness requirement, we show that optimally configured FFR schemes lead to much enhanced performance. We show that the optimally configured directional-FFR schemes increase the throughput capacity by a factor of about 60% relative to that obtained by using optimal omnidirectional-FFR schemes. The analyses and simulation results presented in this paper serve to characterize the performance behavior attainable by using such small cell deployed cellular wireless network systems when employing the underlying configurations.
Summary
TET2 inactivating mutations serve as initiating genetic lesions in the transformation of haematopoietic stem and progenitor cells (HSPCs). In this study, we analysed known drugs in zebrafish ...embryos for their ability to selectively kill tet2‐mutant HSPCs in vivo. We found that the exportin 1 (XPO1) inhibitors, selinexor and eltanexor, selectively kill tet2‐mutant HSPCs. In serial replating colony assays, these small molecules were selectively active in killing murine Tet2‐deficient Lineage‐, Sca1+, Kit+ (LSK) cells, and also TET2‐inactivated human acute myeloid leukaemia (AML) cells. Selective killing of TET2‐mutant HSPCs and human AML cells by these inhibitors was due to increased levels of apoptosis, without evidence of DNA damage based on increased γH2AX expression. The finding that TET2 loss renders HSPCs and AML cells selectively susceptible to cell death induced by XPO1 inhibitors provides preclinical evidence of the selective activity of these drugs, justifying further clinical studies of these small molecules for the treatment of TET2‐mutant haematopoietic malignancies, and to suppress clonal expansion in age‐related TET2‐mutant clonal haematopoiesis.
Here we report the first palladium‐catalyzed asymmetric hydrogenolysis of readily available aryl triflates via desymmetrization and kinetic resolution for facile construction of axially chiral biaryl ...scaffolds with excellent enantioselectivities and s selectivity factors. The axially chiral monophosphine ligands could be prepared from these chiral biaryl compounds and were further applied to palladium‐catalyzed asymmetric allylic alkylation with excellent ee values and high branched and linear ratio, which demonstrated the potential utility of this methodology.
We have developed the first palladium‐catalyzed asymmetric hydrogenolysis of aryl triflates through desymmetrization and kinetic resolution that allows the facile construction of axially chiral biaryl scaffolds with excellent results. These chiral compounds could be further converted into chiral monophosphine ligands, which were then used in asymmetric allylic alkylation to produce the desired product with high regio‐ and enantioselectivities.
Abstract
Ensuring urban areas have access to clean drinking water, safe food supply, and uncontaminated water bodies is essential to the good health of millions of urban residents. This paper ...presents the functionality of Iron Filings-based Green Environmental Media (IFGEM) in terms of nutrient removal efficiencies to support water quality management and urban farming. IFGEM uses recycled materials such as tire crumb and iron filings to help remove nutrients with essential physicochemical properties. In this study, IFGEM were proven effective and sustainable through an isotherm study, a column study of reaction kinetics, and a microstructure examination under various inlet nutrient concentration levels. IFGEMs exhibited over 90% nitrate removal, as well as 50–70% total phosphorus removal, under most inlet conditions. These promising results make IFGEM suitable for treating stormwater runoff, wastewater effluent, and agricultural discharge via varying
ex situ
treatment units in flexible landscape environments. In addition, the byproduct of ammonia generation permits possible reuse of spent IFGEM as soil amendments in crop land, gardens and yards, and green roofs for urban farming. Findings may help secure urban food supply chains and harmonize nutrients, soil, water, and waste management in different urban environments.
Understanding the genetic changes underlying phenotypic variation in sheep (Ovis aries) may facilitate our efforts towards further improvement. Here, we report the deep resequencing of 248 sheep ...including the wild ancestor (O. orientalis), landraces, and improved breeds. We explored the sheep variome and selection signatures. We detected genomic regions harboring genes associated with distinct morphological and agronomic traits, which may be past and potential future targets of domestication, breeding, and selection. Furthermore, we found non-synonymous mutations in a set of plausible candidate genes and significant differences in their allele frequency distributions across breeds. We identified PDGFD as a likely causal gene for fat deposition in the tails of sheep through transcriptome, RT-PCR, qPCR, and Western blot analyses. Our results provide insights into the demographic history of sheep and a valuable genomic resource for future genetic studies and improved genome-assisted breeding of sheep and other domestic animals.