BROWSE
Detail
Conference Paper
Soil Texture Classification using Dual-Depth Soil Moisture Sensor
Cited 0 time in 
Share 
Share
- Authors
- Minjun Kim, Rockwon Kim, Dasong Yu
- Issue Date
- 2023-08
- Citation
- International Conference on Information Technology (ICIT) 2023, pp.200-205
- Language
- English
- Type
- Conference Paper
- Abstract
- This paper represents a preliminary study aimed at introducing a soil texture classification system utilizing data from soil moisture sensors installed in various locations. In this study, we assume major characteristics varying with soil type are the difference in moisture content between the upper and lower layers of the soil and the rate of moisture reduction. To obtain these two features, we use Dual-Depth moisture sensors and define and employ the cumulative moisture decrease equation to acquire the moisture reduction rate. To classify soil type using the data that includes these features, some models based on KNN, AdaBoost, and Random Forest are applied and compared. Additionally, we propose an algorithm that can better handle soil classification based on the Siamese Residual Network (SRN) to not only classify soil types but also to easily compare how similar any given soil is to known soil types. The proposed SRN model got it right about 76% of the time, which is 4% better than the other models.
- KSP Keywords
- Classification system, Preliminary study, Residual Network, Soil Moisture Sensor, Soil classification, Soil texture, Soil types, moisture content(MC), random forest, reduction rate, texture classification