深度学习是一种高效且准确的机器学习方法。由于计算机的快速进步和算法的提

Deep learning is an efficient and accurate machine learning method. Due to the rapid progress of computers and the improvement of algorithms, the development of deep learning has made rapid progress in recent years. <br>The junk image is an image with complex features and diverse categories, which leads to complex extraction and cumbersome recognition. Traditional digital image recognition methods are generally used to identify junk images, which has a large amount of operation and low accuracy. In this paper, a digital image recognition model based on deep learning is used to identify and classify garbage images, and other models are used to compare accuracy and improve the model. The main contents of the paper are: <br>(1) Discuss the relevant theories of deep learning. <br>(2) Discuss related theories of digital image recognition. <br>(3) Study the EfficientNet digital image recognition model based on deep learning, introduce the relevant algorithms and theories of the model, and conduct comparative experiments to improve the model.

深度学习是一种高效且准确的机器学习方法。由于计算机的快速进步和算法的提高，使得近年来深度学习的发展突飞猛进。<br>垃圾图像是一种特征复杂并且类别多样的图像，导致其提取复杂，识别繁冗。一般用传统的数字图像识别方法去识别垃圾图像，运行量大并且准确率低下。本文使用了基于深度学习的数字图像识别模型来对垃圾图像进行识别分类，并使用其他模型进行准确率对比，并对模型进行改进。论文的主要内容有：<br>（1）论述深度学习的相关理论。<br>（2）论述数字图像识别的相关理论。<br>（3）研究基于深度学习的EfficientNet数字图像识别模型，介绍模型的相关算法和相关理论，并进行对比实验，改进模型。

Deep learning is an efficient and accurate machine learning method. Due to the rapid progress of computer and the improvement of algorithm, the development of deep learning has made rapid progress in recent years.<br>Garbage image is a kind of image with complex features and various categories, which leads to complex extraction and recognition. Generally, the traditional digital image recognition method is used to recognize the garbage image, which has a large amount of operation and low accuracy. In this paper, the digital image recognition model based on deep learning is used to classify the garbage image, and the accuracy of other models is compared, and the model is improved. The main contents of this paper are as follows:<br>(1) Discuss the theory of deep learning.<br>(2) This paper discusses the theory of digital image recognition.<br>(3) This paper studies the efficientnet digital image recognition model based on deep learning, introduces the related algorithm and theory of the model, and carries out comparative experiments to improve the model.