Disease Detection in Banana Leaf Plants using DenseNet and Inception Method

Diseases that attack banana plants can affect the growth and productivity of fruit produced. The disease be identified by looking at changes in pattern color leaves. Infected leaves will experience an increased transpiration process photosynthesis is almost non-existent. Furthermore, on cause yield losses up to 50%. Therefore, early detection needed so diseases overcome as soon possible using deep learning. This study aims compare performance DenseNet Inception methods detecting a transfer learning architecture model with fewer parameters computations achieve good performance. Inception, other hand, architectural applies cross-channel correlation, executes lower resolution inputs, avoids spatial dimensions. In conducting test, this uses several data handling schemes test two methods, namely without handling, under-sampling, oversampling. separated into training ratio 80:20. result method oversampling scheme superior models percentage value 84.73% accuracy, 84.80% precision, recall, 84.62% f1 score. addition, machine all also method.