A novel deep LeNet-5 convolutional neural network model for image recognition

Abstract

At present, the traditional machine learning methods and convolutional neural network (CNN) methods are mostly used in image recognition. The feature extraction process in traditional machine learning for image recognition is mostly executed by manual, and its generalization ability is not strong enough. The earliest convolutional neural network also has many defects, such as high hardware requirements, large training sample size, long training time, slow convergence speed and low accuracy. To solve the above problems, this paper proposes a novel deep LeNet-5 convolutional neural network model for image recognition. On the basis of Lenet-5 model with the guaranteed recognition rate, the network structure is simplified and the training speed is improved. Meanwhile, we modify the Logarithmic Rectified Linear Unit (L_ReLU) of the activation function. Finally, the experiments are carried out on the MINIST character library to verify the improved network structure. The recognition ability of the network structure in different parameter s is analyzed compared with the state-of-the-art recognition algorithms. In terms of the recognition rate, the proposed method has exceeded 98%. The results show that the accuracy of the proposed structure is significantly higher than that of the other recognition algorithms, which provides a new reference for the current image recognition.

Performance Evaluation of Optical Transmission Based on Link Estimation by Using Deep Learning Techniques

“…Each selected ar-chitecture contributes unique strengths suitable for the regression nature of the task. The LeNet architecture, being simple and computationally efficient, offers fast processing for image-based regression problems [37]. The Wide ResNet architecture, known for handling varying scales and complexities within images, as discussed in [38], can be crucial for capturing the nuanced changes in optical links.…”

Software System for Automatic Grading of Paper Tests

“…For the purpose of digit recognition, a modified version of the convolutional neural network LENET5 [18] is utilized [19]. The original neural network model exhibits issues of significant bias and variance, and suitable optimizations have been implemented to address these challenges effectively [20][21][22].…”

An Improved Deeplabv3+ Model for Semantic Segmentation of Urban Environments Targeting Autonomous Driving

“…2D semantic segmentation assigns a semantic label to each pixel in an image. The deep learning is also widely used in image recognition [4], image classification [5,6,7,8],object detection [9,10,11,12,13]. In literature [14,15],probability models such as Markov Random Field (MRF) and Conditional Random Field (CRF) are used for semantic segmentation.In addition, with the development of the Convolutional Neural Network (CNN), several studies have used CNN to solve semantic segmentation problems and achieved significant performance improvements.…”