An Adaptive Deep Learning Framework for Prediction of Agricultural Yield

Agriculture and its large yields are vital to human survival. Recently, machine learning and deep learning have been applied to determine agricultural output. This study estimated apple production utilizing a unique deep-learning(DL)-based technique. A study-designed convolutional neural network (CNN) classified golden and royal gala apple photos. Classifying CNN-extracted pictures with traditional machine-learning(ML) methods was a hybrid application. Datasets A is used to evaluate the performance of the proposed model. Based on Dataset A accuracy findings, the CNN-SVM hybrid model performed best. CNN-SVM and CNN-Gradient Boosting hybrid models performed best and accurately for Dataset A. The more balanced distribution of train, test, and validation sizes, the suggested hybrid CNN model, and the model's assessment made dataset C the best choice. The hybrid model was 99.70% accurate for Dataset A. Precision, recall, f-measure, and Cohen kappa were 99%. The study found that hybrid models could accurately predict apple fruit yield using golden and royal gala photos.