Prediction and Detection of Harvesting Stage in Cotton Fields Using Deep Adversarial Networks

Cotton is a crucial crop that has a significant impact on the global economy, and the timing of the harvest is crucial for maximizing the yield and quality of cotton fiber. However, predicting and detecting the harvesting stage of cotton plants is a complex task that requires analyzing various factors such as plant growth, leaf senescence, and boll maturity. Traditional methods for harvesting prediction are labor intensive and time consuming, making it essential to develop efficient and accurate methods. In this paper, we present a novel deep adversarial network (DAN) called CropCycleNet, which combines the features of both convolutional neural networks and generative adversarial networks. The proposed DAN can identify different stages of cotton plant growth, detect diseases, and affect plants to ensure proper removal. We propose Histogram base Gradients Feature Orientation Transform method influences feature descriptors and allows feature-level fusion to improve object recognition accuracy. Experimental validation of CropCycleNet was performed to evaluate the accuracy, precision, recall, and F1 performance metrics at various stages of cotton plant growth. The proposed DAN identified the harvesting stage in cotton fields with 93.27% prediction accuracy, outperforming other existing state-of-the-art methods.