ABSTRACT
Deep learning (DL) is an innovative technology of machine learning (ML); it is a subset of artificial intelligence (AI). Since neural networks (NN) mimic the activities of the human brain, DL will perform the same task. In DL, nothing is explicitly programmed. It is an ML class that uses many nonlinear processing units to perform feature extraction as well as transformation. The production of every preceding layer is engaged as the input of every succeeding layer. DL models are capable enough to focus on the exact features themselves that require less guidance from the programmer and are very helpful in solving dimensional problems. Smart farming in agriculture is a new idea that makes agriculture a productive and structured method that inherits the new technologies. Nowadays, DL involves various research activities that are used to help the farmers to lessen the fatalities in farming. This technology is used to predict which crop is suitable for which weather conditions. Right now, DL, PC vision, picture handling, advanced mechanics, and Internet of Things (IoT) innovations are extremely useful to farmers. Computer-based intelligence controlled drone innovation is extremely valuable for horticulture as it gives top notch pictures to make it more straightforward to screen, examine, and break down crops. This method is helpful for deciding the advancement of a harvest. Additionally, the farmer can conclude regardless of whether the harvest is ready for yield. There is 154no restriction to the portrayal of DL involvement in agribusiness. The study examines typical agricultural uses of deep learning. DL algorithms of convolutional neural networks (CNNs), recurrent neural networks (RNNs), long short-term memory (LSTM) networks, and generative adversarial networks (GANs), radial basis function networks (RBFNs), multi-layer perceptrons (MLPs), self-organizing map (SOM), deep belief networks (DBNs), and restricted Boltzmann machines (RBMs) have been extensively premeditated and are useful in numerous areas of agriculture. Specifically, the applications of CNN-based supervised learning (CNN-SL) and transfer learning advance the development of modern agriculture.
