Results 11 to 20 of about 2,396 (171)
Among all stereo matching methods End‐to‐End (E2E) learning methods show that they can achieve the lowest and most frequent minimum errors, however, their performance drastically changes across different test‐sites, which indicates poor generalisation capabilities.
Hessah Albanwan, Rongjun Qin
wiley +1 more source
A workflow is proposed for a completely automatic orientation of historical terrestrial urban images. For such images, automatic SfM packages often fail due to large radiometric and geometric differences causing challenges with feature extraction and reliable matching.
Ferdinand Maiwald, Hans‐Gerd Maas
wiley +1 more source
Fractional order research has interdisciplinary characteristics and has been widely used in the field of natural sciences. Therefore, fractional order research has become an important area of concern for scholars. This paper used 2854 literatures collected from China National Knowledge Infrastructure (CNKI) database from 2001 to 2020 as the data source
Yunfei Yang +4 more
wiley +1 more source
Applying deep learning to right whale photo identification
Abstract Photo identification is an important tool for estimating abundance and monitoring population trends over time. However, manually matching photographs to known individuals is time‐consuming. Motivated by recent developments in image recognition, we hosted a data science challenge on the crowdsourcing platform Kaggle to automate the ...
Robert Bogucki +5 more
wiley +1 more source
目的 本国际临床指南由欧洲残疾儿童学会(the European Academy of Childhood Disability,EACD)牵头制定,旨在解决发育性协调障碍(developmental coordination disorder,DCD)的定义、诊断、评估、干预以及与社会心理方面的临床应用关键问题。 方法 本指南针对五个领域的关键问题,通过文献综述和专家团队的正式讨论达成共识。为保证指南的循证基础,以“机制”、“评估”和“干预”为检索词, 对2012年更新以来提出的最新建议以及新增的“社会心理问题”和“青少年/成人”为检索词进行检索。根据牛津大学循证医学中心证据等级 (证据水平 [level of evidence, LOE]1–4) 将结果进行分类,最终转化为指南建议。并由国际 ...
Jing Hua +6 more
wiley +1 more source
基于卷积神经网络的网络流量分类算法中,为了提高分类准确度,其结构设计日趋复杂,容易出现梯度下滑甚至梯度消失,导致预测准确度不升反降。文章提出了一种基于残差网络的改进流量分类算法,引入残差网络层代替传统卷积神经网络中的卷积层和池化层,不仅缓解了传统卷积网络因层次太深导致难以训练的问题,同时与传统卷积运算相比,所提出的残差网络在训练时学习到的数据特征信息更加全面,训练后的模型也更加准确。仿真结果表明,改进后的算法比常规的神经网络算法表现更佳,分类准确度从92.05%提高到了96.18%。
陆煜斌 +5 more
doaj
Optimization Technology of Longshan Millet γ-amino Butyric Acid Enrichment Process by Neural Network Algorithm [PDF]
In order to promote the quality and efficiency improvement of Longshan millet industry and enhance the added value of products, this study took Longshan millet as the main raw material to learn the effects of soaking time, germination temperature and ...
GAO Ling +3 more
core +1 more source
针对变电站巡检机器人在传统运动规划方法下存在的难以规划出平滑路径、不确定环境下动作不可测等问题,提出研究不确定条件下的变电站巡检机器人运动规划问题的深度强化学习方法。文中分析了深度学习中奖励值模型II、探索策略和神经网络结构对整个运动规划的影响,设计了不同结构的神经网络,并开展了相关的对比实验。结果表明,在当前任务场景下,相同的运算量神经网络结构C2比神经网络结构C1和神经网络结构C3的计算时间要短。因此,在计算资源短缺时,建议采用神经网络结构C2,更有利于对变电站巡检机器人进行精准的运动规划 ...
董诗绘, 牛彩雯, 戴琨
doaj
Application Study of the Stock Data Assistant Analysis Algorithm [PDF]
股票市场受政治、经济和投资者心理等多种复杂因素的影响。虽然复杂,但经过分析简化仍能在海量数据中得到相应的信息。数据挖掘技术就是能够实现从海量数据中挖掘有用信息的新兴技术,其广泛应用于投资领域。本文以技术分析和基本分析两种股票分析方法为基础,针对股票分析和预测应用数据挖掘中的神经网络和关联规则建立相应的模型,并进行了实证分析和比较。论文主要工作包括以下两个方面: 1.针对BP神经网络算法在股市预测中存在的学习速度慢、容易陷入局部极小值、预测结果精度不高等问题,提出一种基于共轭梯度的BP神经网络算法 ...
魏朝东
core
Computer vision model for the detection of canine pododermatitis and neoplasia of the paw
Background – Artificial intelligence (AI) has been used successfully in human dermatology. AI utilises convolutional neural networks (CNN) to accomplish tasks such as image classification, object detection and segmentation, facilitating early diagnosis.
Andrew Smith +7 more
wiley +1 more source

