基于卷积神经网络的动物图像识别研究

Research on animal image recognition based on convolutional neural network

目前机器学习是计算机科学领域的热点研究问题,机器学习希望计算机能像人一样可以“自主学习”,让机器能基于人类的思维来认识世界,并且能解决人们生产生活中遇到的问题。卷积神经网络是机器学习领域一种十分重要的神经网络结构。图像相较于委婉含蓄的文字更加直观,也更加容易携带丰富的信息,是应用非常广泛的一类信息。图像识别是目前计算机视觉领域应用十分广阔的、代表性的研究技术。而卷积神经网络因为局部感受野的引入、卷积的处理等特点,在图像的识别与处理方面得到了广泛的应用。本文在研究学习了机器学习和CNN的相关框架、理论、算法、应用场景等知识之后,主要进行CNN模型在动物图像识别方向上的实践研究,并在基于Keras框架上设计搭建模型并最终实现对猫狗图像的预测模型。

At present,machine learning is a hot research problem in the field of computer science.Machine learning hopes that computers can“learn independently”like people,so that machines can understand the world based on human thinking,and can solve the problems encountered in people’s production and life.Convolutional neural network is an important neural network structure in the field of machine learning.Compared with euphemistic and implicit words,images are more intuitive and easier to carry rich information,which is a kind of information widely used.Image recognition is a widely used and representative research technology in the field of computer vision.Convolutional neural networks have been widely used in image recognition and processing because of the introduction of local receptive field and the processing of convolution.After studying and learning the relevant framework,theory,algorithm and application scenario of machine learning and CNN,this paper mainly conducted practical research on the CNN model in the direction of animal image recognition,and designed and built a model based on Keras framework and finally realized the prediction model for cat and dog images.