Design and Parameterization of a Computational Model Based on Convolutional Neural Networks for Mango Selection

Published: Feb. 5, 2024

In recent times, using image-based deep learning techniques has been very useful for classifying fruits. However, mango is a particular challenge because its shelf life directly related to the degree of maturity at time harvest, which determined mainly through visual assessment operators when sorting them, doing it incorrectly, coupled with blows and compressions exposed during storage transport, trigger deterioration in their quality, hence importance this operation. sense, purpose study focuses on designing computational model based Deep Learning that uses convolutional neural network (CNN) parameterized select mangoes according color tone stage. A dataset was established from 201 labeled images, then CNN architecture built, obtaining 3 criteria were evaluated: number convolution grouping layers, resolution input images training epochs required. As result, an image 32 x pixels achieved, process included layer followed by pooling fed another processed results ended output 2 neurons under One Hot Encoding representation. The required 20 perfect training, achieving 96.04 % accuracy selection. This supports potential CNNs quality analysis selection, even uncertainty scenarios.

Language: Английский

Drug Delivery via Lipid-Based Nanocarriers

Biomedical materials for multi-functional applications, Journal Year: 2024, Volume and Issue: unknown, P. 297 - 328

Published: Jan. 1, 2024

Language: Английский

Study on the Hepatotoxicity of Emodin and Its Application in the Treatment of Liver Fibrosis

Molecules, Journal Year: 2024, Volume and Issue: 29(21), P. 5122 - 5122

Published: Oct. 30, 2024

Emodin (EMO) is an anthraquinone compound derived from

Language: Английский