In the agriculture field, early detection and diagnosis of plant disease is very critical for the sustainability of crops and food security. Conventional methods such as real-time inspection by experts is hectic, time consuming, and may or may not be factually accurate subject to knowledge of expert(s). The proposed study offers a strategic method to automate the detection and diagnosis process of mango leaf diseases which utilizes the capabilities of ML techniques to address the issue, which undergo preprocessing before being fed into the CNN model for disease prediction. The model’s predictions encompass various mango leaf diseases, including anthracnose, cutting weevil, die back, healthy, and powdery mildew, along with associated confidence levels. The dataset for this research includes photos from four mango orchards in Bangladesh, which contain total of 2,213 samples A comprehensive system evaluation demonstrates the effectiveness of the automated approach in accurately identifying mango leaf diseases with the accuracy rate of 99.83%. By reducing the dependence on human expertise and streamlining disease identification processes, the proposed system empowers farmers to intervene promptly and implement mitigation measures. This crop disease management system holds significant potential to revolutionize agricultural practices, offering valuable insights to farmers and ultimately contributing to enhanced crop productivity and food security.