JoWUA

Abstract

India's economy is supported by crops, which is one of the reasons why it is considered to be one of the developing nations in the globe. Nutrient deficiency in plant management is a challenge for quaponics, which integrates hydroponics with aquaculture to provide sustainable farming techniques. Therefore, to effectively address this matter, developed a hybrid recognition model to enhance the learning outcomes of Deep CNN. This was accomplished by combining the models, namely the VGG- 16 Net and Xception model, into a single model along with explainable deep neural networks (XAI). The ensemble method provides reliable feature extraction and classification by utilizing the advantages of the VGG and Xception models. Researchers provide visual explanations of the detection process by integrating Gradient-weighted Class Activation Maps (Grad-CAM) with the CNN models to improve interpretability and transparency. Detecting Nutrient deficiency’s accuracy is increased by this hybrid approach, which also provides system operators with useful data that promotes proactive management of plant growth. The results validate the ability of the proposed ensemble to accurately diagnose a variety of nutrient deficiencies while maintaining high interpretability. The practice of XAI techniques ensures that the decision-making process remains transparent, fostering reliance and enabling more effective intervention strategies. This research highlights the latent of merging advanced deep learning models with XAI tools to elevate plant health monitoring in aquaponic systems, an actual way for more robust and justifiable agricultural performance.