We present a memory and computation efficient ternary weight networks (TWNs) - with weights constrained to +1, 0 -1. The Euclidian distance between full (float or double) precision the along scaling factor is minimized in training stage. Besides, threshold-based function optimized get an approximated solution which can be fast easily computed. TWNs have shown better expressive abilities than binary counterparts. Meanwhile, achieve up 16$\times$ model compression rate need fewer...

Cross-modality person re-identification (cm-ReID) is a challenging but key technology for intelligent video analysis. Existing works mainly focus on learning modality-shared representation by embedding different modalities into same feature space, lowering the upper bound of distinctiveness. In this paper, we tackle above limitation proposing novel cross-modality shared-specific transfer algorithm (termed cm-SSFT) to explore potential both information and modality-specific characteristics...

Synthesizing photo-realistic images from text descriptions is a challenging problem. Previous studies have shown remarkable progresses on visual quality of the generated images. In this paper, we consider semantics input in helping render However, diverse linguistic expressions pose challenges extracting consistent even they depict same thing. To end, propose novel text-to-image generation model that implicitly disentangles to both fulfill high-level semantic consistency and low-level...

We present a memory and computation efficient ternary weight networks (TWNs) - with weights constrained to +1, 0 -1. The Euclidian distance between full (float or double) precision the along scaling factor is minimized in training stage. Besides, threshold-based function optimized get an approximated solution which can be fast easily computed. TWNs have shown better expressive abilities than binary counterparts. Meanwhile, achieve up 16× model compression rate need fewer multiplications...

In this work, a pH-responsively controlled-release chlorpyrifos (PRCRC) was developed using nanosystem consisting of (CPF), polydopamine (PDA), attapulgite (ATP), and calcium alginate (CA). Therein, CPF adsorbed in the nanonetwork-structured PDA-modified ATP (PA) to obtain CPF-PA through hydrogen bonds electrostatic attraction. Subsequently, combined with CA form porous CPF-PA-CA hydrogel spheres (actually PRCRC) cross-linking reaction, wherein PA acted as skeleton. PRCRC tended collapse...

With the promising applications in e-Health and entertainment services, wireless body area network (WBAN) has attracted significant interest. One critical challenge for WBAN is to track maintain quality of service (QoS), e.g., delivery probability latency, under dynamic environment dictated by human mobility. Another important issue ensure energy efficiency within such a resource-constrained network. In this paper, new medium access control (MAC) protocol proposed tackle these two...

Visible-infrared cross modality person re-identification (VI-ReID) is a core but challenging technology in the 24-hours intelligent surveillance system. How to eliminate large gap lies heart of VI-ReID. Conventional methods mainly focus on directly aligning heterogeneous modalities into same space. However, due unbalanced color information between visible and infrared images, features images tend overfit clothing information, which would be harmful alignment. Besides, these align feature...

Abstract Computer‐aided classification of pathological images is the great significance for breast cancer diagnosis. In recent years, deep learning methods image have made breakthrough progress, becoming mainstream in this field. To capture more discriminant features images, work introduces a novel attention high‐order network (AHoNet) by simultaneously embedding mechanism and statistical representation into residual convolutional network. AHoNet firstly employs an efficient channel module...

Co-occurrent visual pattern makes aggregating contextual information a common paradigm to enhance the pixel representation for semantic image segmentation. The existing approaches focus on modeling context from perspective of whole image, i.e., image-level information. Despite impressive, these methods weaken significance representations same category, semantic-level To address this, this paper proposes augment by and information, respectively. First, an module is designed capture each in...

Controlled release of pesticides by light regulation is one the most viable strategies recently developed for highly efficient utilization agrochemicals. Herein, we report an infrared-light-responsive pesticide delivery system controlled imidacloprid (IMI) preparation functional hollow carbon microspheres (HCMs). After IMI loading and surface functionalization with polyethylene glycol (PEG) α-cyclodextrin (α-CD), was sequestered in (denoted as HCMs/IMI/PEG/α-CD) a result formation PEG/α-CD...

A chiral gold(I) complex-catalyzed highly regio- and enantioselective azo hetero-Diels-Alder reaction has been developed. The complex acting as a Lewis acid exhibits high efficiency in the activation of urea-based diazene dienophiles. Moreover, this gold catalyst also rendered cascade intramolecular enyne cycloisomerization/asymmetric azo-HDA reaction.

Automatic segmentation of brain tumors from magnetic resonance imaging (MRI) is a challenging task due to the uneven, irregular and unstructured size shape tumors. Recently, tumor methods based on symmetric U-Net architecture have achieved favorable performance. Meanwhile, effectiveness enhancing local responses for feature extraction restoration has also been shown in recent works, which may encourage better performance problem. Inspired by this, we try introduce attention mechanism into...

Dense captioning aims at simultaneously localizing semantic regions and describing these regions-of-interest (ROIs) with short phrases or sentences in natural language. Previous studies have shown remarkable progresses, but they are often vulnerable to the aperture problem that a caption generated by features inside one ROI lacks contextual coherence its surrounding context input image. In this work, we investigate reasoning based on multi-scale message propagations from neighboring contents...

Ship detection technology is an important development direction in the field of optical remote sensing image processing. In recent years, convolutional neural networks have achieved good results ship target and recognition. We train latest model YOLOv5 on our dataset this paper. The show that can be well applied detection.

On-body wireless networks (oBWNs) play a crucial role in improving the ubiquitous healthcare services. Using oBWNs, vital physiological information of patient can be gathered from wearable sensor nodes and accessed by authorized user like health professional or doctor. Since open nature communication sensitivity information, secure has always been issue oBWNs-based systems. In recent years, several authentication schemes have proposed for remote monitoring. However, most these are so...

Wireless body area network (WBAN) is an emerging technology that provides socialized health monitoring service. However, the quality of service can be severely degraded by concomitant inter-WBAN interference in some specific environments where multiple WBANs are densely deployed, e.g., hospitals and senior citizen communities. In this work, we propose a Bayesian game based power control scheme to mitigate impact interference. By modeling as players active links types model, proposed tries...

Recent online multi-object tracking (MOT) methods have achieved desirable performance. However, the speed of most existing is rather slow. Inspired from fact that adjacent frames are highly relevant and redundant, we divide into key non-key track objects in compressed domain. For frames, RGB images restored for detection data association. To make association more reliable, an appearance convolutional neural network (CNN) which can be jointly trained with detector proposed. directly...

Many of the recent methods for semi-supervised video object segmentation are still far from being applicable real time applications due to their slow inference speed. Therefore, we explore a propagation based method in compressed domain accelerate speed this paper. In particular, only extract features I-frames by traditional deep convolutional neural network and produce P-frames through information flow propagation. process feature propagation, propose two effective components enhance...