An improved LeNet-5 model for Image Recognition
Authors:
Open Access
- 6 November 2020
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 444-448
Abstract
This paper proposes an improved convolutional neural network structure which greatly reduces the scale of network training parameters. It uses the global average pooling algorithm instead of the full connection algorithm to improve the LeNet-5 network. The number of convolution kernels is increased, while the number of subsampling layers is reduced to an optimal value. After verification with MINST handwritten Arabic numeral data set, the results show that the improved network training parameters are only 34.8% of the original, and the recognition accuracy can achieve 99.3%.Keywords
This publication has 5 references indexed in Scilit:
- Study on Traffic Sign Recognition by Optimized Lenet-5 AlgorithmInternational Journal of Pattern Recognition and Artificial Intelligence, 2019
- The image-based analysis and classification of urine sediments using a LeNet-5 neural networkComputer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization, 2019
- Development of a LeNet-5 Gas Identification CNN Structure for Electronic NosesSensors, 2019
- Masked face detection via a modified LeNetNeurocomputing, 2016
- Gradient-based learning applied to document recognitionProceedings of the IEEE, 1998