Investigating Efficacy of Deep Trained Soil Classification System with Augmented Data

Farmers need to be aware of the correct soil type for a specific crop to maximize agricultural yield, which affects the rising demand for food. In this paper, an appropriate and efficient soil classification system was aimed to propose by implementing deep learning approaches. Image-based soil data set was collected and pre-processed according to algorithmic requirements. Initially, classification was implemented using machine learning classification algorithms and then it compares with deep learning algorithms. Due to fewer images approximately 30 images in five categories, algorithmic training was resulted in low. To improve accuracy data augmentation was implemented. Further, the augmented dataset was utilized to train the machine learning and deep learning models. Based on the comparison, efficient algorithms were proposed.