– Soil organic matter (SOM) and soil moisture contents (SMC) are two main properties in defining soil health. It is a challenge to measure organic matter and moisture content in soil, as the conventional methods are time, labor and money consuming. In order to overcome these challenges, various image processing based models have been proposed to predict SOM and SMC. Proposed model uses a stepwise multiple linear regression (SMLR) method to predict these properties on basis of soil color features like color moments, GLCMs and different color models as well, since, soil moisture and organic content are influenced by soil color. Multiple soil samples from field are collected with a certain distance in order to simulate continuous variation in the contents. Loss of ignition method is used to generate the ground truth to feed the model. For produce and compare the result first 34 and then 6 optimal predictor variables are used in model. The output results in external validation for SOM prediction were: R 2 = 0.07, RMSE = 0.76, RPIQ = 1.00 and that for SMC were: R 2 = 0.77, RMSE = 0.55, RPIQ = 1.07.

Soil texture is one of the crucial characteristic in determining soil health. Classifying soil texture manually 1 is expensive, time consuming and requires experienced experts who are often limited available. Multiple machine leaning algorithms are proposed in the recent past to hold up a fully automated soil texture classification in 12 or lesser classes using soil images. Among such algorithms research on deep neural networks (DNNs) has been explored less. Wherever these DNNs are applied, they are used in isolation. Limited efforts are made to transfer the knowledge from 5 DNN of some other application and reuse the pre-trained network. In this work, concept of transfer learning is 6 investigated in soil texture prediction. Inceptionv3, ResNet50 and ResNet152 are trained on soil image dataset which 7 consists of images acquired from agricultural fields of multiple crops. It also shows analysis of different image 8 processing based segmentation techniques.