Classical-Quantum Computing Model with MobileNet for Precise Pest Classification
T. Saranya, C. Deisy, S. Sridevi · 2023
Pest classification holds immense significance in agriculture, serving as a cornerstone in safeguarding crop health and global food security. This paper introduces a hybrid classical- quantum machine learning model with the objective of revolutionizing pest classification. The proposed methodology embraces the Classical-Quantum (CQ) paradigm, amalgamating quantum circuit design, the Quantum-Enhanced Classical Network (QCN), and seamless integration with the MobileNetV2architecture. Remarkably, this hybrid model outperforms MobileNetV2 in terms of accuracy, boasting an exceptional training accuracy of 98% and an impressive testing accuracy of 96%. These achievements are realized on a dataset encompassing 4263 images of tomato-specific pests categorized into 8 distinct classes. The CQ model attains this remarkable performance by harnessing the power of quantum computing to extract more distinctive features from pest images. The QCN component of the model facilitates efficient and scalable training of quantum- enhanced classical classifiers, contributing to its superior performance. The integration of the MobileNetV2 architecture provides the model with a lightweight and efficient architecture that is suitable for deployment on mobile devices.