Cassava Disease Classification with Mobittention Architecture

Early detection in agriculture boosts crop resilience, minimizes yield losses, and supports timely interference against diseases and pests. Deep earning (DL) and Convolutional Neural Networks (CNNs) outshine in pattern analysis by learning features from pixel data. Existing models have difficulty in correctly detecting and classifying cassava diseases, therefore an efficient modern deep learning is proposed in this paper. In this paper, a Mobittention Model is proposed where the MobileNet architecture is combined with Efficient Channel Attention (ECA) modules to boost detection performance. The proposed model is tested on the Cassava Leaf Disease dataset. The result shows that our model outperforms the existing state of art methods in terms of accuracy.