CONFERENCE / ICCAIS-2026

Banana plant life are crucial for meals protection in tropical regions, regardless of the reality that bananas are susceptible to diseases inclusive of leaf spot sickness (LSD), which can seriously affect crops. Traditional detection strategies are labor-intensive and errors-inclined, while present computerized systems often lack sufficient accuracy in unique environments, leaving a big hollow in illness manipulate. This look at develops a Convolutional Neural Network (CNN) model based totally on ResNet50 shape to categorise banana leaf illnesses into four lessons Cordana, Sigatoka, Healthy and Pestalotiopsis. The version makes use of statistical upgrades to simulate one of a kind field situation, enhancing its generalization abilties. Preliminary consequences display that the ResNet50-based definitely model achieves promising accuracy in the course of the disease vicinity, suggesting its capacity as a precious device for early detection. This automated reaction may be covered into mobile gadgets, facilitating proactive illness control and helping sustainable banana farming practices.