Intensification of Agriculture using Deep Learning and Machine Learning: A Survey

Agriculture has been a significant aspect in developing countries where most of the rural population depends on it for their livelihoods. Hence, making sure of the health and yield of the crop is something really called for. There are many reasons due to which growth of crops gets affected and hence its yield. Some of the salient factors are diseases, environmental conditions like temperature, humidity, etc., and the presence of weeds. In this paper, a study of these factors has been done. Initially, a study of crop disease detection with an in-depth study of segmentation and feature extraction is done. Diverse techniques, like the Support Vector Machine (SVM), Convolutional Neural Network (CNN), AlexNet for disease detection. Genetic algorithms, Otsu algorithm, Random Forest for segmentation, are compared to determine the most effective one for the desired system. In the next part of the study, weed classification algorithms like CNN and SVM are studied and a thorough comparison is done. The crop yield can be affected by different environmental factors like temperature, surface reflectance, and humidity. Different algorithms like CNN, LSTM (Long Short - Term Memory), and the CNN-LSTM framework, that serves the need of crop yield prediction, are assessed and conclusions are drawn.