Classifying soil texture images using transfer learning

International Conference on Information Technology and Digital Applications 2018

计算机科学;无线电电子学

Guidang, E.P.B.^1

Abra State Institute of Sciences and Technology, Lagangilang, Abra, Philippines^1

关键词: Feature mining;    Literature reviews;    Machine learning techniques;    Soil textures;    Trained neural networks;    Transfer learning;    Transfer learning methods;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/482/1/012042/pdf
DOI  :  10.1088/1757-899X/482/1/012042

学科分类：计算机科学（综合）

来源: IOP

【 摘 要 】

Transfer learning is a machine learning technique which makes use of a pre-trained neural network to classify new objects. This study was conducted to evaluate the performance of Inception-v3 in classifying Soil texture images on different conditions. Specifically it achieved the following objectives 1) Identified the features of Inception-v3 and 2) Classified Soil Texture images using Inception-v3. The study used literature review to identify the features of Inception-v3. The study found that Transfer Learning comprises of two portions: a) feature mining and b) classification. Moreover, Inception-v3 highest prediction rating of a Soil texture image is 98% and 86% as the lowest. The study concludes that Transfer Learning method through the use of Inception-v3 can be used to classify Soil texture images.