To solve the problems of the traditional convolutional neuralnetwork's needs of long training time and poor accuracy in the process of fruit image classification, the present study proposes a fruit image classifi...
详细信息
To solve the problems of the traditional convolutional neuralnetwork's needs of long training time and poor accuracy in the process of fruit image classification, the present study proposes a fruit image classification method based on the multi-optimization convolutional neuralnetwork with the background of fruit classification. Firstly, in order to avoid the interference of external noise and influence the accuracy of classification, the wavelet threshold is used to denoise the fruit image, which can reduce image noise while preserving the details of the image. Secondly, to correct the over-bright fruit image or the over-dark fruit image, the gamma transform is adopted to correct the image. Finally, in the process of constructing the convolutional neuralnetwork, the SOM network is introduced for pre-learning the samples. Besides, the weights of the trained optimal SOM network are applied to the full connection layer, and an integrated optimization model of convolution and full connection is established for feature extraction and regression classification. The optimized convolutional neuralnetwork was adopted to classify fruits. According to the application results, the accuracy of the optimized convolutional neuralnetwork for fruit classification reaches 99%. Therefore, the improved convolutional neuralnetwork depth learning algorithm makes better performance to achieve fruit classification.
When dealing with high-dimensional multivariate time series classification problems, a well-known difficulty is the curse of dimensionality. In this article, we propose an original approach of transposition of multidi...
详细信息
ISBN:
(纸本)9781728198354
When dealing with high-dimensional multivariate time series classification problems, a well-known difficulty is the curse of dimensionality. In this article, we propose an original approach of transposition of multidimensional data into images to tackle the task of classification. We propose a lightweight hybrid model that take this transposed data as an input. This model contains convolutional layers as a feature extractor followed by a recurrent neuralnetwork. We apply our method to a large dataset consisting of individual patient medical records. We show that our approach allows us to significantly reduce the size of a network and increase its performance by opting for a transformation of the input data.
in this work, we present an end-to-end system devoted to automatic recognition of printed Amazigh script in complex document images containing different languages such as Web images and natural scene images. To this e...
详细信息
ISBN:
(纸本)9781538605516
in this work, we present an end-to-end system devoted to automatic recognition of printed Amazigh script in complex document images containing different languages such as Web images and natural scene images. To this end, text extraction from images is performed;the extracted text serves as input for a trained convolutional neuralnetwork (CNN) to identify its language. Finally, we proceed to the recognition of the Amazigh text script using a developed optical character recognition (OCR) system. The CNN reaches 99,12% of accuracy while the OCR system gets 99,93% The obtained results seem to be very satisfactory.
Potato is one of India's main agricultural crops. In recent years, potato farms have become increasingly popular in India. However, a variety of ailments are driving up farmers' costs in the production of pota...
详细信息
暂无评论