Reformer: Re-parameterized kernel lightweight transformer for grape disease segmentation DOI
Weisong Mu,

Zibo Feng,

Weisong Mu

и другие.

Expert Systems with Applications, Год журнала: 2024, Номер 265, С. 125757 - 125757

Опубликована: Дек. 9, 2024

Язык: Английский

A systematic review of deep learning techniques for plant diseases DOI Creative Commons
İshak Paçal, İsmail Kunduracıoğlu, Mehmet Hakkı Alma

и другие.

Artificial Intelligence Review, Год журнала: 2024, Номер 57(11)

Опубликована: Сен. 30, 2024

Язык: Английский

Процитировано

25
MFHSformer: Hierarchical sparse transformer based on multi-feature fusion for soil pore segmentation DOI

Hao Bai,

Qiaoling Han,

Yandong Zhao

и другие.

Expert Systems with Applications, Год журнала: 2025, Номер 272, С. 126789 - 126789

Опубликована: Фев. 7, 2025

Язык: Английский

Процитировано

1
FATDNet: A fusion adversarial network for tomato leaf disease segmentation under complex backgrounds DOI

Zaichun Yang,

Lixiang Sun, Zhihuan Liu

и другие.

Computers and Electronics in Agriculture, Год журнала: 2025, Номер 234, С. 110270 - 110270

Опубликована: Март 20, 2025

Язык: Английский

Процитировано

1
Recent advances in Transformer technology for agriculture: A comprehensive survey DOI
Weijun Xie,

M G Zhao,

Ying Liu

и другие.

Engineering Applications of Artificial Intelligence, Год журнала: 2024, Номер 138, С. 109412 - 109412

Опубликована: Окт. 11, 2024

Язык: Английский

Процитировано

6
Low-Cost Lettuce Height Measurement Based on Depth Vision and Lightweight Instance Segmentation Model DOI Creative Commons

Yiqiu Zhao,

Xiaodong Zhang,

Jingjing Sun

и другие.

Agriculture, Год журнала: 2024, Номер 14(9), С. 1596 - 1596

Опубликована: Сен. 13, 2024

Plant height is a crucial indicator of crop growth. Rapid measurement facilitates the implementation and management planting strategies, ensuring optimal production quality yield. This paper presents low-cost method for rapid multiple lettuce heights, developed using an improved YOLOv8n-seg model stacking characteristics planes in depth images. First, we designed lightweight instance segmentation based on by enhancing architecture reconstructing channel dimension distribution. was trained small-sample dataset augmented through random transformations. Secondly, proposed to detect segment horizontal plane. leverages plane, as identified image histogram from overhead perspective, allowing identification parallel camera’s imaging Subsequently, evaluated distance between each plane centers contours select cultivation substrate reference bottom height. Finally, plants determined calculating difference top plant. The experimental results demonstrated that achieved 25.56% increase processing speed, along with 2.4% enhancement mean average precision compared original model. accuracy plant algorithm reached 94.339% hydroponics 91.22% pot scenarios, absolute errors 7.39 mm 9.23 mm, similar sensor’s direction error. With images downsampled factor 1/8, highest speed recorded 6.99 frames per second (fps), enabling system process 174 targets second. confirmed exhibits promising accuracy, efficiency, robustness.

Язык: Английский

Процитировано

5
Natural disaster damage analysis using lightweight spatial feature aggregated deep learning model DOI
Kibitok Abraham, Mohammed Abo‐Zahhad,

Moataz M. Abdelwahab

и другие.

Earth Science Informatics, Год журнала: 2024, Номер 17(4), С. 3149 - 3161

Опубликована: Май 28, 2024

Язык: Английский

Процитировано

4
PlanText: Gradually Masked Guidance to Align Image Phenotype with Trait Description for Plant Disease Texts DOI Creative Commons

Kejun Zhao,

Xingcai Wu,

Yuanyuan Xiao

и другие.

Plant Phenomics, Год журнала: 2024, Номер 6

Опубликована: Янв. 1, 2024

Plant diseases are a critical driver of the global food crisis. The integration advanced artificial intelligence technologies can substantially enhance plant disease diagnostics. However, current methods for early and complex detection remain challenging. Employing multimodal technologies, akin to medical diagnostics that combine diverse data types, may offer more effective solution. Presently, reliance on single-modal predominates in research, which limits scope detailed diagnosis. Consequently, developing text modality generation techniques is essential overcoming limitations recognition. To this end, we propose method aligning phenotypes with trait descriptions, diagnoses by progressively masking images. First, training validation, annotate 5,728 phenotype images expert diagnostic provide annotated labels 210,000 Then, PhenoTrait description model, consists heterogeneous feature encoders as well switching-attention decoders, accurate context-aware output. Next, generate phenotypically appropriate description, adopt 3 stages embedding image features into semantic structures, characterizations preserve features. Finally, our experimental results show model outperforms several frontier models multiple including larger GPT-4 GPT-4o. Our code dataset available at https://plantext.samlab.cn/.

Язык: Английский

Процитировано

3
A deep learning-based micro-CT image analysis pipeline for nondestructive quantification of the maize kernel internal structure DOI Creative Commons
Juan Wang, Si Yang,

Chuanyu Wang

и другие.

Plant Phenomics, Год журнала: 2025, Номер unknown, С. 100022 - 100022

Опубликована: Фев. 1, 2025

Язык: Английский

Процитировано

0
CMAA: Channel-wise multi-scale adaptive attention network for metallographic image semantic segmentation DOI

Yongliang Sun,

Xiangyang Huang

Expert Systems with Applications, Год журнала: 2025, Номер unknown, С. 126925 - 126925

Опубликована: Март 1, 2025

Язык: Английский

Процитировано

0
Robust crop disease detection using Multi-Domain Data Augmentation and Isolated Test-Time Adaptation DOI Creative Commons
Rui Fu, Han Jiao,

Yumei Sun

и другие.

Expert Systems with Applications, Год журнала: 2025, Номер unknown, С. 127324 - 127324

Опубликована: Апрель 1, 2025

Язык: Английский

Процитировано

0