Improving plant disease classification by adaptive minimal ensembling

Abstract

A novel method for improving plant disease classification, a challenging and time-consuming process, is proposed. First, using as baseline EfficientNet, a recent and advanced family of architectures having an excellent accuracy/complexity trade-off, we have introduced, devised, and applied refined techniques based on transfer learning, regularization, stratification, weighted metrics, and advanced optimizers in order to achieve improved performance. Then, we go further by introducing adaptive minimal ensembling, which is a unique input to the knowledge base of the proposed solution. This represents a leap forward since it allows improving the accuracy with limited complexity using only two EfficientNet-b0 weak models, performing ensembling on feature vectors by a trainable layer instead of classic aggregation on outputs. To the best of our knowledge, such an approach to ensembling has never been used before in literature. Our method was tested on PlantVillage, a public reference dataset used for benchmarking models' performances for crop disease diagnostic, considering both its original and augmented versions. We noticeably improved the state of the art by achieving 100% accuracy in both the original and augmented datasets. Results were obtained using PyTorch to train, test, and validate the models; reproducibility is granted by providing exhaustive details, including hyperparameters used in the experimentation. A Web interface is also made publicly available to test the proposed methods.

Efficient Deep Learning Approach for Olive Disease Classification

“…Data is organized best-to-worst fold (top-to-bottom), and then the models corresponding to the first two rows in the left table are used as weak models for the ensemble. This approach to ensembling has been recently introduced and discussed in [21], [22]; it has already proved excellent applicability to AI-based methods for agriculture [23].…”

GranoScan: an AI-powered mobile app for in-field identification of biotic threats of wheat

“…To obtain a classification system for the images we collected, we opted to use an original method that we studied and implemented. More in detail, to get accurate and robust AI models, we used a deep learning architecture named efficient minimal adaptive ensembling that we already tested ( Bruno et al., 2022 ) by setting the new state-of-the-art with an accuracy of 100% on the Plantvillage public dataset. The method is based on an ensembling strategy that uses as core models two instances of the EfficientNet-b0 architecture.…”

OLID I: an open leaf image dataset for plant stress recognition

“…The PlantVillage dataset has consistently served as the benchmark in this context. Nevertheless, researchers recently achieved 100% accuracy on it ( Bruno et al., 2022 ). We aspired to offer a successor to PlantVillage with additional stressor categories encompassing hitherto unexplored crops and high-resolution photographs.…”