Classification of Banana Plant Disease Using Hybrid CNN

One important aspect of Indias agricultural economy is banana farming. Still farmers frequently face a challenge as a result of the crops many diseases. Early detection of these infections is essential to avoiding financial losses. Due to a few issues the current hybrid deep learning model which combines SVM and DWT performs less accurately. Though SVM is a very strong technique it is still unusual to achieve this amazing level of accuracy. Discrete Wavelet Transform (DWT) drawbacks include sensitivity to shifts, non-stationary signal challenges, boundary effects, lack of shift invariance, and limited time-frequency localization. CNNs however may perform worse and be more prone to overfitting in situations where labelled data is scarce. A hybrid approach combining mobile net and densenet has been developed to detect and classify banana disease in order to solve this issue. By suggesting suitable fertilizers to stop diseases in their tracks this creative method helps farmers. Compared to other deep learning techniques in this field the proposed method has demonstrated an impressive accuracy rate of 98.7%.