Abstract The detection of tiny defects in industrial products is important for improving the quality and maintaining production safety. Currently, image-based defect methods are ineffective detecting variously shaped defects. Therefore, this paper proposes a network (TD-Net) to improve effectiveness detection. TD-Net improves overall effect, especially effect defects, by solving problems downsampling pre-filtering conflicting deep shallow semantic information, cascading fusion multi-scale...

Abstract Our study aims to understand the impact of open government data (OGD) policy on firm performance and moderating role characteristics. The difference‐in‐difference model is used analyze 10‐year panel from 477 Chinese listed firms. findings indicate that (a) OGD positively affects performance; (b) size moderates relationship between (c) R&D intensity nonlinearly (d) significantly state‐owned This paper provides suggestions for improving market construction.

Aiming at the problems of low efficiency, high false detection rate, and poor real-time performance current industrial defect methods, this paper proposes an method based on expanded perceptual field feature fusion for practical applications. First, to improve network, original network structure is enhanced by using depth-separable convolution reduce computation while ensuring accuracy, critical information extraction from map MECA (More Efficient Channel Attention) attention network. To...

Road surface defect detection can effectively reduce maintenance costs, which is a critical component in road structural health monitoring. However, existing methods often face challenges the heavy computational parameters and high-accuracy detection, limiting their practical applicability resource-constrained industrial settings. To alleviate this gap, we propose Lightweight High-accuracy Network (LHA-Net) for consisting of three sub-networks feature extraction, fusion, head. First,...

Detecting product surface defects is an important issue in industrial scenarios. In the actual scene, shooting angle and distance between camera object often vary, which results a large variation scale angle. addition, high-speed cameras are prone to motion blur, further deteriorates defect detection results. order solve above problems, this study proposes model for products based on attention enhancement. The network takes advantage of lower-level higher-resolution feature map from backbone...

Abstract Accurate power load prediction is an important guide for system planning and operation. High‐ or low‐load results will affect the operation of system. In recent years, deep learning technology represented by convolution neural network (CNN) transformer has been proved to be suitable prediction. This paper proposes a new short‐term hybrid forecasting model, called channel enhanced attention (CEA) temporal convolutional (TCN)‐based comprehensive model. method combines feature...

Open government data (OGD) are considered a sustainable driver of firm innovation. Leadership is crucial decision-maker for firms to employ OGD in The present study focuses on two the most prominent leadership styles: transformational and transactional. Drawing Organizational Commitment Theory, we claim that affective normative commitment parallel mechanisms explain how promotes OGD-driven innovation firms. Our results show radical through commitment. In contrast, transactional incremental...

The pothole is a common road defect that seriously affects traffic efficiency and personal safety. Road evaluation maintenance automatic driving take detection as their main research part. In the above scenarios, accuracy real-time are most important. However, current methods can not meet requirements of due to multiple parameters volume. To solve these problems, we first propose lightweight one-stage object network, AAL-Net. design an LF (lightweight feature extraction) module use NAM...

Effective log anomaly detection can help operators locate and solve problems quickly, ensure the rapid recovery of system, reduce economic losses. However, recent studies have shown some drawbacks, such as concept drift, noise problems, fuzzy feature relation extraction, which cause data instability abnormal misjudgment, leading to significant performance degradation. This paper proposes a multi-feature deep fusion an unstable model (MDFULog) for above problems. The MDFULog uses novel...

Defect detection of industrial products is crucial to ensure product quality and use safety. However, the following difficulties still exist complex defect background, large similarity between classes weak semantic information, extensive image noise information. In order solve this challenge above, paper proposes a novel Interactive Convolutional Transformer-based Encoder-Decoder Detection Network (ICT-EDNet). Specifically, ICT-EDNet has three characteristics. First, designs edge-interactive...

Industrial quality detection is one of the important fields in machine vision. Big data analysis, Internet Things, edge computing, and other technologies are widely used industrial detection. Studying an algorithm that can be organically combined with Things computing imminent. Deep learning methods have been proposed recently. However, due to particularity scenarios, existing deep learning-based general object shortcomings applications. This study designs two isomorphic models solve these...

Formula recognition is widely used in document intelligent processing, which can significantly shorten the time for mathematical formula input, but accuracy of traditional methods could be higher. In order to solve complexity an end-to-end encoder-decoder framework with attention mechanism proposed that converts formulas pictures into LaTeX sequences. The Vision Transformer (VIT) employed as encoder convert original input picture a set semantic vectors. Due two-dimensional nature formula,...

To prevent the compilation of documents, many table documents are formatted with non-editable and non-structured texts such as PDFs or images. Quickly recognizing contents tables is still a challenge due to factors irregular formats, uneven text quality, complex diverse content. This article proposes UTTSR recognition model, which consists four parts: region detection, line detection recognition, sequence recognition. For Cascade Faster RCNN ResNeXt105 network implemented, using TPS (Thin...

Relation classification is an important fundamental task in information extraction, and convolutional neural networks have been commonly applied to relation with good results. In recent years, due the proposed pre-training model BERT, use of which as a feature extraction architecture has become more popular, gradually withdrawn from stage NLP, classification/extraction based on BERT achieved state-of-the-art However, none these methods consider how accurately capture semantic features...

The big data industry is a major opportunity to promote the development of digital economy. This paper analyzes status quo and problems in Shandong Province, combines connotation chain, puts forward countermeasure suggestions for