Hybrid ensemble framework combining hard and soft voting of deep transfer learning models

Abstract

Plant is the main source of producing food for humans. In agriculture, to increase plant production with good quality and large quantity, farmers are continuously making efforts day and night. Multiple reasons can affect crop production. The proposed system focuses on early disease detection in plants. Farmers need the help of agricultural experts to recognize the disease on the plant. However, it leads to an economically costly and time-consuming process for both the farmer and the agriculturist. For automatic detection of diseases on plants, tremendous work has been done using different technologies like image processing, computer vision, Internet of Things, and Artificial Intelligence, with excellent performance. The proposed work has used a transfer learning approach, a type of artificial intelligence, and the Cucumber dataset with multiple diseases. Hybrid soft and hybrid hard voting two different approaches of ensemble learning are used with transfer learning models for the evaluation. Hybrid soft voting evaluated similar accuracy to an individual model that is strong accuracy transfer learning trained model, but inference time is much less than that of these strong models. Therefore, hybrid soft voting is useful for the fast prediction of diseases.

Abstract from the paper’s OpenAlex record, shown for indexing and research. Copyright remains with the authors or the publisher. Read at the source ↗

The text the model read, with the source record’s inline markup (italics, sub- and superscripts) rendered and its character entities decoded; quote offsets refer to the raw text.