当前位置: X-MOL 学术IETE Tech. Rev. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Effective Processing of Convolutional Neural Networks for Computer Vision: A Tutorial and Survey
IETE Technical Review ( IF 2.5 ) Pub Date : 2020-09-27 , DOI: 10.1080/02564602.2020.1823252
Ronald Tombe 1 , Serestina Viriri 1
Affiliation  

Over the past few years, Convolutional neural networks (ConvNets) is emerging as computer vision discipline within deep learning. ConvNets is a key strategy for addressing computer vision problems, yet the theories behind their effectiveness in the processing are not yet fully understood. ConvNets have attained a state of the art-performances on various datasets for computer vision tasks such as remote sensing images scene classification, face recognition, and object detection. This is attributed to their effectiveness in image feature processing. This work reviews the major advances on ConvNets for effective processing in computer vision from some dimensions which include, convolutional layer design configurations, pooling layer strategies, network activation functions, loss functions, network regularization techniques, and ConvNet optimization methods. Further, this works surveys the application of ConvNets on three computer vision tasks, i.e. remote sensing images scene recognition, face recognition, and object detection to demonstrate the effectiveness of convNets in image feature processing. Additionally, datasets for evaluation and benchmarking purposes with the aforementioned computer vision tasks are briefly discussed.



中文翻译:

用于计算机视觉的卷积神经网络的有效处理:教程和调查

在过去的几年里,卷积神经网络 (ConvNets) 正在成为深度学习中的计算机视觉学科。ConvNets 是解决计算机视觉问题的关键策略,但其处理有效性背后的理论尚未完全理解。ConvNets 在用于计算机视觉任务(如遥感图像场景分类、人脸识别和对象检测)的各种数据集上取得了最先进的性能。这归因于它们在图像特征处理中的有效性。这项工作从卷积层设计配置、池化层策略、网络激活函数、损失函数、网络正则化技术、和 ConvNet 优化方法。此外,这项工作调查了 ConvNets 在三个计算机视觉任务中的应用,遥感图像场景识别、人脸识别和物体检测,以证明卷积神经网络在图像特征处理中的有效性。此外,还简要讨论了用于上述计算机视觉任务的评估和基准测试的数据集。

更新日期:2020-09-27
down
wechat
bug