The ICDAR 2019 Challenge on "Scanned receipts OCR and key information extraction" (SROIE) covers important aspects related to the automated analysis of scanned receipts. SROIE tasks play a role in many document systems hold significant commercial potential. Although lot work has been published over years administrative analysis, community advanced relatively slowly, as most datasets have kept private. One contributions is offer first, standardized dataset 1000 whole receipt images...

In this paper, we introduce a novel end-end framework for multi-oriented scene text detection from an instance-aware semantic segmentation perspective. We present Fused Text Segmentation Networks, which combine multi-level features during the feature extracting as instance may rely on finer expression compared to general objects. It detects and segments jointly simultaneously, leveraging merits both task region proposal based object task. Not involving any extra pipelines, our approach...

Driven by the vision of Internet Things, some research efforts have already focused on designing a network efficient speech recognition for development edge computing. Other researches (such as tpool2) do not make full use spatial and temporal information in acoustic features speech. In this paper, we propose compact with spatio-temporal computing, named EdgeRNN. Alternatively, EdgeRNN uses 1-Dimensional Convolutional Neural Network (1-D CNN) to process overall each frequency domain...

This article studies the robust fuzzy attitude control strategy for networked spacecraft consisting of physically independent modules that interact through network communication. The effects inertia uncertainties, delayed feedback, actuator saturation, and external disturbances are simultaneously taken into account when designing law. Based on aperiodic sampling, an adaptive event-triggered scheme avoiding Zeno phenomenon naturally is introduced to reduce communication burden network. By...

Recurrent neural network (RNN) has been widely applied to many sequential tagging tasks such as natural language process (NLP) and time series analysis, it proved that RNN works well in those areas. In this paper, we propose using with long short-term memory (LSTM) units for server load performance prediction. Classical methods prediction focus on building relation between domain, which makes a lot of unrealistic hypotheses. Our model is built based events (user requests), the root cause...

Cluster assignment of large and complex datasets is a crucial but challenging task in pattern recognition computer vision. In this study, we explore the possibility employing fuzzy clustering deep neural network framework. Thus, present novel evolutionary unsupervised learning representation model with iterative optimization. It implements adaptive (DAFC) strategy that learns convolutional classifier from given only unlabeled data samples. DAFC consists feature quality-verifying model, where...

In this paper we present a research on identification of audio recording devices from background noise, thus providing method for forensics. The signal is the sum speech and noise signal. Usually, people pay more attention to signal, because it carries information deliver. So great amount researches have been dedicated getting higher Signal-Noise-Ratio (SNR). There are many enhancement algorithms improve quality speech, which can be seen as reducing noise. However, noises regarded intrinsic...

Recognizing text in images has been a hot research topic computer vision for decades due to its various application. However, the variations appearance term of perspective distortion, line curvature, styles, etc., cause great trouble recognition. Inspired by Transformer structure [1] that achieved outstanding performance many natural language processing related applications, we propose new Transformer-like recognition images, which is referred as Hierarchical Attention Network (HATN). The...

Cloud computing has been exploited in managing large-scale IoT systems. cloud servers usually handle a large number of requests from various devices. Due to the fluctuant and heavy workload, require provide high scalability, stable performance, low price necessary functionalities. However, traditional clouds offer service with abstraction virtual machine (VM), which can hardly meet these requirements. Meanwhile, different vendors performance stabilities models, fluctuate according dynamic...

As for cluster-based wireless sensor networks (WSNs), cluster lifetime is one of the most important subjects in recent researches. Besides reducing energy consumptions clusters, it necessary to make clusters achieve equal lifetimes so that whole network can survive longer. In this paper, we focus on multi-hop WSNs with cooperative multi-input single-output scheme. With a simplified model WSNs, change transmission schemes, sizes and distances investigate their effects lifetimes. Furthermore,...

In this paper, we introduce a novel end-end framework for multi-oriented scene text detection from an instance-aware semantic segmentation perspective. We present Fused Text Segmentation Networks, which combine multi-level features during the feature extracting as instance may rely on finer expression compared to general objects. It detects and segments jointly simultaneously, leveraging merits both task region proposal based object task. Not involving any extra pipelines, our approach...

Mis- and disinformation online have become a major societal problem as sources of harms different kinds. One common form mis- is out-of-context (OOC) information, where pieces information are falsely associated, e.g., real image combined with false textual caption or misleading description. Although some past studies attempted to defend against OOC through external evidence, they tend disregard the role evidence stances. Motivated by intuition that stance represents bias towards detection...