Fish diseases impact food security and result in financial losses in the aquaculture sector. Conventional disease detection strategies are defined by time, knowledge, and possible inapplicability to large-scale operations. In response to these challenges, with an eye toward MobileNetV2, we developed a deep learning–based automated system for diagnosing fish diseases. Trained on a properly chosen dataset, a convolutional neural network (CNN) filters fish images into disease categories. Images of both sick and healthy fish are abundant, illustrating the conditions in an aquaculture facility. Image resizing, normalizing, and augmenting help one improve generalizability and model resilience. The effective FDDM (MobileNetV2) model helps to smooth down the generated dataset; it performs well on cellphones with limited processing capability. Model parameters are evaluated during training using the following metrics: accuracy (99.71%), precision (99.7%), recall (99%), and F1-score (99%), which help to reduce classification errors. The model's ability to identify diseases in fish suggests its potential utility in real-time disease surveillance in aquaculture. Early identification of fish diseases made possible by this scalable and reasonably priced approach helps lower the demand for human inspections, enabling a quick reaction. It will help improve aquaculture sustainability, fish health management, and product quality.