Plant crop disease detection is becoming more crucial nowadays due to the demand in enable automation of industrial performance. In order to tackle this type of disease, we introduce ResNet50, a deep learning (DL) model for the classification of tomato crop plant images by Convolutional Neural Networks (CNNs). This applied model distinguishes between healthy leaf and TB-disease-affected leaf along with assessing the extent of disease severity. Second, a ternary classification system diagnoses whether a plant is healthy or diseased. In summary, the method is able to meet the needs of 5 different binary ratings for severity into 99.3% accuracy compared 92% for multi-class ratings. This involves the use of technology or digital solutions in agriculture field to provide better, earlier detection of leaf disease that is necessary for improving crop yield. Our approach using the reputable and publicly available Plant Village Dataset achieves high efficiency for improving tomato crop health management. This development highlights technology is making positive effects on agriculture practices in current farming, it helps farmers to improve their crop health management, and monitoring, and finally benefits consumer through this process.