English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Machine vision can provide a convenient and intelligent recognition for tomato leaf diseases. However, limited labelled images affect the performance of such a method to a considerable extent. This study proposes a dual-strategy semi-supervised learning method for recognizing tomato leaf diseases when labelled images are insufficient. One of the dual strategies attaches an additional classifier  C  to the generative adversarial network (GAN) to construct a semi-supervised learning method (GANC) which generates artificial images to tune itself based on the labelled images with the label removal, and the high confidence pseudo-label comes from the classifier prediction. For the other strategy, the discriminator of the pretrained GAN continues to modify the output as authenticity and class probability (D GAN ), and unlabelled images are integrated to optimize the network. Then, the predicted results of GANC and D GAN  are combined to construct the dual-strategy semi-supervised learning method (GANC – D GAN ). GANC could generate new unlabelled images with a similar data distribution as real disease images and obtain an accuracy of 84.06% in the prediction set (ACC p ), and D GAN  utilizes exiting unlabelled images to obtain an ACC P  with 85.31%, outperforming supervised learning with a small number of labelled images. With image augmentation, the ACC p  of GANC and D GAN  increases to 88.13% and 90.00%, respectively. GANC – D GAN  can achieve an accuracy of 90.62%, which is better than those of the two individual strategies as GANC and D GAN . Thus, the proposed dual-strategy semi-supervised learning method could offer accurate disease recognition even if the available labelled images are insufficient.

Dual-strategy semi-supervised learning method based on GAN for recognition of tomato leaf diseases