【计算机科学】【2018.02】【含源码】一种目标分类的深度学习预测模型

本文为荷兰代尔夫特理工大学（作者：N.E. Sahla）的硕士论文，共58页。

在过去的十年，全球仓储自动化市场迅速而显著地增长。最大的挑战在于识别和处理不同的物件。本研究旨在探讨物件特征，例如大小或形状与条形码位置之间是否存在可用的关系，从而稳健地辨识储存箱中的物件。在MATLAB中建立了一个深度卷积神经网络（CNN），并在一个由来自不同角度的数千幅产品图像组成的标记数据集上进行训练，以确定条形码位于产品的哪个表面。训练结果表明，训练数据集精度达到100%，而验证数据的最大精度只有45%。需要更大的数据集以减少过度拟合，并提高验证精度。当达到足够的分类精度时，智能拣选策略可以被有效地用于实际产品的处理。

The last decade has marked a rapid andsignificant growth of the global market of warehouse automation. The biggestchallenge lies in the identification and handling of foreign objects. The aimof this research is to investigate whether a usable relation exist betweenobject features such as size or shape, and barcode location, that can be usedto robustly identify objects in a bin. A deep convolutional neural network(CNN) is built in MATLAB and trained on a labeled dataset of thousand productimages from various perspectives, to determine on which surface of a productthe barcode lies. Training results show that while the training set accuracy reaches100%, a maximum validation accuracy of only 45% is achieved. A larger datasetis required to reduce overfitting and increase the validation accuracy. Whensufficient classification accuracies are reached, smart picking strategies canbe implemented to efficiently handle products.

1 引言

2 机器学习

3 机器学习算法回顾

4 目标辨识理论框架

5 深度学习的实现

6 结论

附录 MATLAB代码：卷积神经网络

下载英文原文地址：

http://page5.dfpan.com/fs/0lcbj2f21e296161357/

更多精彩文章请关注微信号：