Open Set Self and Across Domain Adaptation for Tomato Disease Recognition With Deep Learning Techniques

Abstract

Recent advances in automatic recognition systems based on deep learning technology have shown the potential to provide environmental-friendly plant disease monitoring. These systems are able to reliably distinguish plant anomalies under varying environmental conditions as the basis for plant intervention using methods such as classification or detection. However, they often show a performance decay when applied under new field conditions and unseen data. Therefore, in this article, we propose an approach based on the concept of open-set domain adaptation to the task of plant disease recognition to allow existing systems to operate in new environments with unseen conditions and farms. Our system specifically copes diagnosis as an open set learning problem, and mainly operates in the target domain by exploiting a precise estimation of unknown data while maintaining the performance of the known classes. The main framework consists of two modules based on deep learning that perform bounding box detection and open set self and across domain adaptation. The detector is built based on our previous filter bank architecture for plant diseases recognition and enforces domain adaptation from the source to the target domain, by constraining data to be classified as one of the target classes or labeled as unknown otherwise. We perform an extensive evaluation on our tomato plant diseases dataset with three different domain farms, which indicates that our approach can efficiently cope with changes of new field environments during field-testing and observe consistent gains from explicit modeling of unseen data.

From Laboratory to Field: Unsupervised Domain Adaptation for Plant Disease Recognition in the Wild

“… Visualizations of the learned representations using t-SNE on the C-PD dataset for Baseline [ 44 ], DAAN [ 33 ], D-Coral [ 42 ], DAN [ 17 ], DANN [ 43 ], MRAN [ 18 ], DSAN [ 27 ], and our MSUN, where subfigure (A) is colored based on the plant species, while subfigure (B) is colored based on the datasets. …”

“…Although being widely studied in the community of computer vision owing to its tremendous advantages, research in domain adaptation on plant disease classification is rare. We only found 2 related studies [ 13 , 18 ]. Yan et al [ 13 ] propose a cross-species plant disease recognition framework using UDA, where a mixed subdomain alignment method is employed to solve the multiclassification task of plant disease severity.…”

“…Yan et al [ 13 ] propose a cross-species plant disease recognition framework using UDA, where a mixed subdomain alignment method is employed to solve the multiclassification task of plant disease severity. Fuentes et al [ 18 ] propose domain adaption for tomato disease classification, which can cope with changes of the environments. However, both of these 2 methods directly apply the existing UDA to classify plant diseases, thus failing to consider some issues exclusively existing in various plant diseases of cross species.…”

Data-centric annotation analysis for plant disease detection: Strategy, consistency, and performance

“…The dashed box is the focus of our research. Dong et al 10.3389/fpls.2022.1037655 2020; Fuentes et al, 2021;Fenu and Malloci, 2021) ignored the impact of the annotation strategy on the detection model. In addition, image classification systems recently made a giant leap with the advancement of deep neural networks, which require sufficient accurate labeled data to be adequately trained.…”

Distributed Averaging Problems of Agriculture Picking Multi-Robot Systems via Sampled Control

“…Recently, the issue of smart agriculture has attracted wide attention. Driven by the digital revolution, agriculture has entered a new era of digital and intelligent development (Yang et al, 2013 ; Alsamhi et al, 2019a ; Horng et al, 2019 ; Fuentes et al, 2021 ; Li and Chao, 2022 ; Teng et al, 2022 ). Smart agriculture is a modern agricultural production mode with information, theoretical knowledge, and hardware equipment as the core elements, and it is an important direction of the development of modern agriculture (Chen and Yang, 2019 ).…”