A Comparative Analysis on Transfer Learning Models to Classify Banana Diseases- Fusarium Wilt and Black Sigatoka

The global appeal of bananas, both as a fruit and a vegetable, is undeniable due to their versatility. However, the productivity of banana cultivation faces significant threats from fungal diseases, notably Fusarium wilt and Black Sigatoka, which can devastate entire plantations. This study introduces a deep learning-based methodology to automate the detection of these diseases, employing a comparative analysis of several algorithms: VGG19, NasNetLarge, Inception Resnet V2, and Xception. Our findings reveal that the Xception model outperforms others, achieving an accuracy of 98.062%, with precision, recall, and F1score all at 99%. Implementing machine-driven early detection systems could prove critical in safeguarding the livelihoods of banana farmers against such pervasive threats.