Learning to Synergize Semantic and Geometric Priors for Limited-Data Wheat Disease Segmentation

Scott Chapman; Shijie Wang; Xin Yu; Yadan Luo; Zi Huang; Zijian Wang

arxiv: 2604.05415 · v1 · submitted 2026-04-07 · 💻 cs.CV

Learning to Synergize Semantic and Geometric Priors for Limited-Data Wheat Disease Segmentation

Shijie Wang , Zijian Wang , Yadan Luo , Scott Chapman , Xin Yu , Zi Huang This is my paper

Pith reviewed 2026-05-10 19:16 UTC · model grok-4.3

classification 💻 cs.CV

keywords wheat disease segmentationlimited datasemantic priorsgeometric priorspoint promptsappearance invarianceprecision agricultureadapter tuning

0 comments

The pith

A framework turns pretrained semantic features into filtered point prompts that activate geometric boundary localization for accurate wheat disease segmentation despite limited training data and large appearance shifts.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper aims to show that wheat disease segmentation can be reframed as the joint problem of semantic perception that tolerates temporal variations and precise geometric boundary detection. By inserting disease-specific adapters into existing pretrained models and converting semantic features into dense category-specific point prompts, the method supplies comprehensive spatial coverage. These prompts are then dynamically filtered using mask confidence scores cross-checked against semantic consistency, which activates the geometric model's strengths while removing redundancy. If successful, this synergy would let practitioners achieve high-quality segmentations without collecting the large representative datasets that temporal appearance changes normally demand. The result is segmentation performance that stays stable across growth stages where standard training from scratch struggles.

Core claim

The central claim is that pretrained semantic priors provide category-aware robustness to intra-class temporal variations and can be transformed into dense point prompts; after dynamic filtering that cross-references mask generation confidence with semantic consistency, these prompts activate geometric priors to yield precise, boundary-accurate disease masks that remain invariant to appearance shifts across growth stages.

What carries the argument

The prompt synergization pipeline that converts semantic features into dense category-specific point prompts and then filters them by iterative mask confidence and semantic consistency to guide precise boundary localization.

If this is right

Delivers state-of-the-art results on wheat disease and organ segmentation benchmarks particularly when annotated data are scarce.
Produces masks whose accuracy does not degrade when disease appearance changes substantially between growth stages.
Reduces the volume of labeled examples required to reach usable segmentation quality in precision agriculture settings.
Ensures comprehensive spatial coverage of all disease regions while suppressing redundant or low-quality prompts.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The same prompt-conversion and filtering logic could be tested on segmentation tasks in other crops that also exhibit strong seasonal or stage-dependent appearance variation.
If the filtering step proves robust, the approach might reduce annotation budgets for related agricultural monitoring problems such as nutrient deficiency or pest damage mapping.
The explicit separation of semantic prompting from geometric localization offers a template for pairing other pretrained model families when labeled data are limited.

Load-bearing premise

That semantic features from a pretrained model can be turned into comprehensive yet non-redundant point prompts whose filtering by mask confidence will correctly retain only accurate candidates across every possible temporal appearance shift.

What would settle it

A new wheat image collection drawn from previously unseen growth stages and disease variants where the method's segmentation accuracy falls below that of standard fine-tuning baselines on the same limited training split.

Figures

Figures reproduced from arXiv: 2604.05415 by Scott Chapman, Shijie Wang, Xin Yu, Yadan Luo, Zi Huang, Zijian Wang.

**Figure 3.** Figure 3: Visualization of point prompts across different sam [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

read the original abstract

Wheat disease segmentation is fundamental to precision agriculture but faces severe challenges from significant intra-class temporal variations across growth stages. Such substantial appearance shifts make collecting a representative dataset for training from scratch both labor-intensive and impractical. To address this, we propose SGPer, a Semantic-Geometric Prior Synergization framework that treats wheat disease segmentation under limited data as a coupled task of disease-specific semantic perception and disease boundary localization. Our core insight is that pretrained DINOv2 provides robust category-aware semantic priors to handle appearance shifts, which can be converted into coarse spatial prompts to guide SAM for the precise localization of disease boundaries. Specifically, SGPer designs disease-sensitive adapters with multiple disease-friendly filters and inserts them into both DINOv2 and SAM to align their pretrained representations with disease-specific characteristics. To operationalize this synergy, SGPer transforms DINOv2-derived features into dense, category-specific point prompts to ensure comprehensive spatial coverage of all disease regions. To subsequently eliminate prompt redundancy and ensure highly accurate mask generation, it dynamically filters these dense candidates by cross-referencing SAM's iterative mask confidence with the category-specific semantic consistency derived from DINOv2. Ultimately, SGPer distills a highly informative set of prompts to activate SAM's geometric priors, achieving precise and robust segmentation that remains strictly invariant to temporal appearance changes. Extensive evaluations demonstrate that SGPer consistently achieves state-of-the-art performance on wheat disease and organ segmentation benchmarks, especially in data-constrained scenarios.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

SGPer gives a concrete adapter-plus-prompt pipeline for limited-data wheat disease segmentation using DINOv2 and SAM, but the performance gains still need the actual tables to be convincing.

read the letter

The paper's main contribution is a specific way to make DINOv2 and SAM work together on wheat diseases when labels are scarce. It adds disease-sensitive adapters to both models, turns DINOv2 features into dense category-specific point prompts, and then filters those prompts by checking SAM mask confidence against DINOv2 semantic consistency. This targets the temporal appearance shifts across growth stages without needing to train everything from scratch. The design choices—multiple filters in the adapters and the cross-reference step—are clearly laid out and feel like a direct response to the practical bottleneck in precision agriculture. That part is useful and new enough on its own terms. The limited-data framing is honest about why collecting representative datasets is hard for this crop. The pipeline description avoids obvious internal contradictions and treats the synergy as an empirical outcome rather than a logical necessity. The soft spot is the performance claim. The abstract says it reaches SOTA on wheat disease and organ benchmarks especially under data constraints, yet the provided text gives no numbers, no baseline comparisons, no ablation results, and no dataset details. Without those, it's impossible to tell whether the adapters and filtering actually move the needle or whether the pretrained models are carrying most of the load. The filtering logic could also add errors if the semantic consistency check fails on certain stages, but that remains untested in what I can see. This paper is for researchers working on few-shot or prompt-based segmentation in agriculture or similar narrow domains. A reader who wants to adapt foundation models to temporal variation in crops would find the implementation steps worth looking at. It deserves a serious referee because the motivation and mechanism are coherent and the problem is real, even if the results section will need close checking. I would send it for review but flag the missing quantitative evidence as the first thing to address.

Referee Report

2 major / 2 minor

Summary. The paper proposes SGPer, a Semantic-Geometric Prior Synergization framework for wheat disease segmentation under limited data. It inserts disease-sensitive adapters into pretrained DINOv2 and SAM, converts DINOv2 features into dense category-specific point prompts for comprehensive coverage, and filters redundant prompts by cross-referencing SAM iterative mask confidence against DINOv2 semantic consistency. The central claim is that this synergy yields state-of-the-art performance on wheat disease and organ segmentation benchmarks, especially in data-constrained settings, while remaining robust to temporal appearance shifts.

Significance. If the empirical results hold, the work could be significant for precision agriculture by showing how to adapt foundation models (DINOv2 for semantics, SAM for geometry) to domain-specific limited-data tasks without full retraining. The prompt-generation-plus-filtering pipeline offers a concrete mechanism for leveraging pretrained priors on problems with high intra-class variation.

major comments (2)

[Methods / Prompt Filtering] The prompt filtering procedure (described after the adapter insertion in the methods) is load-bearing for the robustness claim. The paper asserts that cross-referencing SAM mask confidence with DINOv2 semantic consistency eliminates redundancy without introducing new errors across temporal shifts, yet no ablation isolating the filter (e.g., dense prompts vs. filtered prompts) or failure-case analysis on appearance changes is referenced; this leaves the central performance advantage unsubstantiated.
[Experiments] The SOTA claim in the abstract and evaluation section rests on quantitative evidence that is not supplied in the provided manuscript excerpt. Specific metrics (mIoU, Dice, etc.), baseline comparisons (including vanilla SAM, DINOv2-only, and prior wheat segmentation methods), dataset splits, and data-constraint regimes (e.g., 1-shot, 5-shot) must be presented with statistical significance to support the assertion that SGPer outperforms existing approaches.

minor comments (2)

[Abstract] The abstract is overly dense in its pipeline description; a single sentence summarizing the two-stage prompt generation and filtering would improve readability.
[Methods] Notation for the adapters and filters (e.g., 'disease-friendly filters') should be introduced with a consistent symbol or acronym on first use and reused in the methods section.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. We address each major comment point by point below and outline the revisions we will make to strengthen the presentation of our method and results.

read point-by-point responses

Referee: [Methods / Prompt Filtering] The prompt filtering procedure (described after the adapter insertion in the methods) is load-bearing for the robustness claim. The paper asserts that cross-referencing SAM mask confidence with DINOv2 semantic consistency eliminates redundancy without introducing new errors across temporal shifts, yet no ablation isolating the filter (e.g., dense prompts vs. filtered prompts) or failure-case analysis on appearance changes is referenced; this leaves the central performance advantage unsubstantiated.

Authors: We agree that the prompt filtering step is central to the robustness claims and that its contribution should be isolated. The current manuscript describes the cross-referencing mechanism but does not include a dedicated ablation comparing dense versus filtered prompts or a targeted failure-case analysis under temporal shifts. In the revised manuscript we will add an ablation study in the Experiments section that reports mIoU and Dice scores for both variants across limited-data regimes and temporal appearance changes. We will also add a qualitative failure-case analysis in the supplementary material to show that the filter does not introduce new errors. revision: yes
Referee: [Experiments] The SOTA claim in the abstract and evaluation section rests on quantitative evidence that is not supplied in the provided manuscript excerpt. Specific metrics (mIoU, Dice, etc.), baseline comparisons (including vanilla SAM, DINOv2-only, and prior wheat segmentation methods), dataset splits, and data-constraint regimes (e.g., 1-shot, 5-shot) must be presented with statistical significance to support the assertion that SGPer outperforms existing approaches.

Authors: We acknowledge that the excerpt supplied to the referee does not contain the detailed quantitative results. To substantiate the state-of-the-art claims, the revised manuscript will expand the evaluation section to include all requested elements: mIoU and Dice scores, comparisons against vanilla SAM, DINOv2-only, and prior wheat segmentation methods, explicit dataset split information, results under 1-shot and 5-shot regimes, and statistical significance obtained from multiple independent runs. revision: yes

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The paper proposes SGPer as an empirical framework that converts DINOv2 semantic priors into point prompts for SAM, using newly designed adapters and a cross-referencing filter for redundancy removal. All load-bearing steps are architectural choices (adapters, prompt generation, confidence-based filtering) whose effectiveness is asserted via benchmark evaluations rather than any closed-form equations, fitted parameters renamed as predictions, or self-citation chains. The SOTA claim in limited-data settings is presented as an experimental outcome, not a logical necessity derived from the inputs themselves. No self-definitional loops, ansatzes smuggled via prior work, or uniqueness theorems appear in the provided description.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract provides no explicit free parameters, axioms, or invented entities; the approach rests on the unstated assumption that general-purpose foundation models can be lightly adapted for domain-specific temporal invariance without detailed justification of adapter placement or filter design.

pith-pipeline@v0.9.0 · 5575 in / 1307 out tokens · 49695 ms · 2026-05-10T19:16:01.733343+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

39 extracted references · 39 canonical work pages

[1]

Shubhra Aich and Ian Stavness. 2017. Leaf Counting with Deep Convolutional and Deconvolutional Networks. In2017 IEEE International Conference on Computer Vision Workshops, ICCV Workshops 2017, Venice, Italy, October 22-29, 2017. IEEE Computer Society, 2080–2089

work page 2017
[2]

Mathilde Caron, Hugo Touvron, Ishan Misra, Hervé Jégou, Julien Mairal, Piotr Bojanowski, and Armand Joulin. 2021. Emerging Properties in Self-Supervised Vision Transformers. In2021 IEEE/CVF International Conference on Computer Vision, ICCV 2021, Montreal, QC, Canada, October 10-17, 2021. IEEE, 9630–9640

work page 2021
[3]

Shoufa Chen, Chongjian Ge, Zhan Tong, Jiangliu Wang, Yibing Song, Jue Wang, and Ping Luo. 2022. Adaptformer: Adapting vision transformers for scalable visual recognition.Advances in Neural Information Processing Systems35 (2022), 16664–16678

work page 2022
[4]

Tianrun Chen, Lanyun Zhu, Chaotao Ding, Runlong Cao, Yan Wang, Shangzhan Zhang, Zejian Li, Lingyun Sun, Ying Zang, and Papa Mao. 2023. SAM-Adapter: Adapting Segment Anything in Underperformed Scenes. InIEEE/CVF Interna- tional Conference on Computer Vision, ICCV 2023 - Workshops, Paris, France, October 2-6, 2023. IEEE, 3359–3367

work page 2023
[5]

A ConvNet for the 2020s

Bowen Cheng, Ishan Misra, Alexander G. Schwing, Alexander Kirillov, and Ro- hit Girdhar. 2022. Masked-attention Mask Transformer for Universal Image Segmentation. InIEEE/CVF Conference on Computer Vision and Pattern Recog- nition, CVPR 2022, New Orleans, LA, USA, June 18-24, 2022. IEEE, 1280–1289. doi:10.1109/CVPR52688.2022.00135

work page doi:10.1109/cvpr52688.2022.00135 2022
[6]

Schwing, and Alexander Kirillov

Bowen Cheng, Alexander G. Schwing, and Alexander Kirillov. 2021. Per-Pixel Classification is Not All You Need for Semantic Segmentation. InAdvances in Neu- ral Information Processing Systems 34: Annual Conference on Neural Information Processing Systems 2021, NeurIPS 2021, December 6-14, 2021, virtual, Marc’Aurelio Ranzato, Alina Beygelzimer, Yann N. Daup...

work page 2021
[7]

Etienne David, Simon Madec, Pouria Sadeghi-Tehran, Helge Aasen, Bangyou Zheng, Shouyang Liu, Norbert Kirchgessner, Goro Ishikawa, Koichi Nagasawa, Minhajul A Badhon, et al. 2020. Global wheat head detection (GWHD) dataset: A large and diverse dataset of high-resolution RGB-labelled images to develop and benchmark wheat head detection methods.Plant Phenomics(2020)

work page 2020
[8]

Zehua Fan, Cuiping Liu, Fengyuan Yu, Yongliang Lai, Qiang Wang, Zhiyong Zhang, Juanjuan Zhang, Jinpeng Cheng, Xinming Ma, Xiaohe Gu, and Shuping Xiong. 2026. Dynamic prediction of carbon and nitrogen accumulation in winter wheat grain: Source-sink theory integrated with UAV multispectral imagery. Comput. Electron. Agric.240 (2026), 111228

work page 2026
[9]

Mathieu Pagé Fortin. 2023. Class-Incremental Learning of Plant and Disease Detection: Growing Branches with Knowledge Distillation. InIEEE/CVF Inter- national Conference on Computer Vision, ICCV 2023 - Workshops, Paris, France, October 2-6, 2023. IEEE, 593–603

work page 2023
[10]

Meng-Hao Guo, Cheng-Ze Lu, Qibin Hou, Zhengning Liu, Ming-Ming Cheng, and Shi-Min Hu. 2022. SegNeXt: Rethinking Convolutional Attention De- sign for Semantic Segmentation. InAdvances in Neural Information Process- ing Systems 35: Annual Conference on Neural Information Processing Sys- tems 2022, NeurIPS 2022, New Orleans, LA, USA, November 28 - Decem- ber...

work page 2022
[11]

Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Liang Wang, Weizhu Chen, et al. 2022. Lora: Low-rank adaptation of large language models.Iclr1, 2 (2022), 3

work page 2022
[12]

Jie Jiang, Zhaopeng Fu, Jiayi Zhang, Jinpeng Yang, Wanping Fang, Qiang Cao, Yongchao Tian, Yan Zhu, Weixing Cao, and Xiaojun Liu. 2026. Developing a nitrogen management strategy for winter wheat to enhance economic profit and energy savings using satellite-based management zone mapping at the county scale.Comput. Electron. Agric.244 (2026), 111527

work page 2026
[13]

Lei Ke, Mingqiao Ye, Martin Danelljan, Yifan Liu, Yu-Wing Tai, Chi-Keung Tang, and Fisher Yu. 2023. Segment Anything in High Quality. InAdvances in Neural Information Processing Systems 36: Annual Conference on Neural Information Processing Systems 2023, NeurIPS 2023, New Orleans, LA, USA, December 10 - 16, 2023, Alice Oh, Tristan Naumann, Amir Globerson,...

work page 2023
[14]

Kingma and Jimmy Ba

Diederik P. Kingma and Jimmy Ba. 2015. Adam: A Method for Stochastic Opti- mization. In3rd International Conference on Learning Representations, ICLR 2015, San Diego, CA, USA, May 7-9, 2015, Conference Track Proceedings, Yoshua Bengio and Yann LeCun (Eds.)

work page 2015
[15]

Berg, Wan-Yen Lo, Piotr Dollár, and Ross B

Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloé Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alexander C. Berg, Wan-Yen Lo, Piotr Dollár, and Ross B. Girshick. 2023. Segment Anything. InIEEE/CVF International Conference on Computer Vision, ICCV 2023, Paris, France, October 1-6, 2023. IEEE, 3992–4003

work page 2023
[16]

Yu Lei, Chen Yang, Jinling Zhao, Xue Yang, Chunchun Li, Wenjiang Huang, Hongbo Qiao, and Linsheng Huang. 2026. PPBM-YOLO: A lightweight and accurate detection model for multi-species airborne spores in wheat disease Cross-Infection scenarios.Comput. Electron. Agric.241 (2026), 111325

work page 2026
[17]

Jiachen Li, Jitesh Jain, and Humphrey Shi. 2024. Matting Anything. InIEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2024 - Workshops, Seattle, W A, USA, June 17-18, 2024. IEEE, 1775–1785

work page 2024
[18]

Yang Liu, Muzhi Zhu, Hengtao Li, Hao Chen, Xinlong Wang, and Chunhua Shen

work page
[19]

InThe Twelfth International Conference on Learning Representations, ICLR 2024, Vienna, Austria, May 7-11, 2024

Matcher: Segment Anything with One Shot Using All-Purpose Feature Matching. InThe Twelfth International Conference on Learning Representations, ICLR 2024, Vienna, Austria, May 7-11, 2024. OpenReview.net

work page 2024
[20]

Maxime Oquab, Timothée Darcet, Théo Moutakanni, Huy V. Vo, Marc Szafraniec, Vasil Khalidov, Pierre Fernandez, Daniel Haziza, Francisco Massa, Alaaeldin El-Nouby, Mido Assran, Nicolas Ballas, Wojciech Galuba, Russell Howes, Po- Yao Huang, Shang-Wen Li, Ishan Misra, Michael Rabbat, Vasu Sharma, Gabriel Synnaeve, Hu Xu, Hervé Jégou, Julien Mairal, Patrick La...

work page 2024
[21]

Huihui Pan, Yuanduo Hong, Weichao Sun, and Yisong Jia. 2023. Deep Dual- Resolution Networks for Real-Time and Accurate Semantic Segmentation of Traffic Scenes.IEEE Trans. Intell. Transp. Syst.24, 3 (2023), 3448–3460. doi:10. 1109/TITS.2022.3228042

work page arXiv 2023
[22]

Yongming Rao, Wenliang Zhao, Guangyi Chen, Yansong Tang, Zheng Zhu, Guan Huang, Jie Zhou, and Jiwen Lu. 2022. DenseCLIP: Language-Guided Dense Prediction with Context-Aware Prompting. InIEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2022, New Orleans, LA, USA, June 18-24,

work page 2022
[23]

Vijai Singh and Ak K Misra. 2017. Detection of plant leaf diseases using image seg- mentation and soft computing techniques.Information processing in Agriculture 4, 1 (2017), 41–49

work page 2017
[24]

Yanpeng Sun, Jiahui Chen, Shan Zhang, Xinyu Zhang, Qiang Chen, Gang Zhang, Errui Ding, Jingdong Wang, and Zechao Li. 2024. VRP-SAM: SAM with Vi- sual Reference Prompt. InIEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2024, Seattle, W A, USA, June 16-22, 2024. IEEE, 23565–23574

work page 2024
[25]

Shijie Wang, Zijian Wang, Yadan Luo, Pengfei Zhang, and Xin Yu. [n. d.]. Parameter-Efficient Wheat Disease Segmentation. InDatabases Theory and Appli- cations - 36th Australasian Database Conference, ADC 2025, Sydney, NSW, Australia and Bali, Indonesia, December 4-6, 2025, Proceedings, Renata Borovica-Gajic, Arijit Khan, Bolong Zheng, Xiaoyang Wang, and J...

work page 2025
[26]

Zijian Wang, Radek Zenkl, Latifa Greche, Benoit De Solan, Lucas Bernigaud Samatan, Safaa Ouahid, Andrea Visioni, Carlos A Robles-Zazueta, Francisco Pinto, Ivan Perez-Olivera, et al . 2025. The global wheat full semantic organ segmentation (GWFSS) dataset.Plant Phenomics7, 3 (2025), 100084

work page 2025
[27]

Chapman, and Zi Huang

Tianqi Wei, Xin Yu, Zhi Chen, Scott C. Chapman, and Zi Huang. 2025. Aug- ment to Segment: Tackling Pixel-Level Imbalance in Wheat Disease and Pest Segmentation. InDatabases Theory and Applications - 36th Australasian Database Conference, ADC 2025, Sydney, NSW, Australia and Bali, Indonesia, December 4-6, 2025, Proceedings (Lecture Notes in Computer Scienc...

work page 2025
[28]

Junde Wu, Rao Fu, Huihui Fang, Yuanpei Liu, Zhaowei Wang, Yanwu Xu, Yueming Jin, and Tal Arbel. 2023. Medical SAM Adapter: Adapting Segment Anything Model for Medical Image Segmentation.CoRRabs/2304.12620 (2023)

work page arXiv 2023
[29]

Álvarez, and Ping Luo

Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, José M. Álvarez, and Ping Luo. 2021. SegFormer: Simple and Efficient Design for Seman- tic Segmentation with Transformers. InAdvances in Neural Information Pro- cessing Systems 34: Annual Conference on Neural Information Processing Sys- tems 2021, NeurIPS 2021, December 6-14, 2021, virtual, Marc’Aurelio...

work page 2021
[30]

Mengde Xu, Zheng Zhang, Fangyun Wei, Han Hu, and Xiang Bai. 2023. SAN: Side Adapter Network for Open-Vocabulary Semantic Segmentation.IEEE Trans. Pattern Anal. Mach. Intell.45, 12 (2023), 15546–15561. doi:10.1109/TPAMI.2023. 3311618

work page doi:10.1109/tpami.2023 2023
[31]

Guoyu Yang, Yuan Wang, Daming Shi, and Yanzhong Wang. 2025. Golden Cudgel Network for Real-Time Semantic Segmentation. InIEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2025, Nashville, TN, USA, June 11-15, 2025. Computer Vision Foundation / IEEE, 25367–25376

work page 2025
[32]

Jinyu Yang, Mingqi Gao, Zhe Li, Shang Gao, Fangjing Wang, and Feng Zheng

work page
[33]

Track Anything: Segment Anything Meets Videos.CoRRabs/2304.11968 (2023)

work page arXiv 2023
[34]

Dongshuo Yin, Yiran Yang, Zhechao Wang, Hongfeng Yu, Kaiwen Wei, and Xian Sun. 2023. 1% vs 100%: Parameter-efficient low rank adapter for dense predictions. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 20116–20126

work page 2023
[35]

Wenxi Yue, Jing Zhang, Kun Hu, Yong Xia, Jiebo Luo, and Zhiyong Wang. 2024. SurgicalSAM: Efficient Class Promptable Surgical Instrument Segmentation. In Thirty-Eighth AAAI Conference on Artificial Intelligence, AAAI 2024, Thirty-Sixth Conference acronym ’XX, June 03–05, 2018, Woodstock, NY Wang et al. Conference on Innovative Applications of Artificial In...

work page 2024
[36]

Radek Zenkl, Radu Timofte, Norbert Kirchgessner, Lukas Roth, Andreas Hund, Luc Van Gool, Achim Walter, and Helge Aasen. 2022. Outdoor plant segmentation with deep learning for high-throughput field phenotyping on a diverse wheat dataset.Frontiers in plant science12 (2022), 774068

work page 2022
[37]

Renrui Zhang, Zhengkai Jiang, Ziyu Guo, Shilin Yan, Junting Pan, Hao Dong, Yu Qiao, Peng Gao, and Hongsheng Li. 2024. Personalize Segment Anything Model with One Shot. InThe Twelfth International Conference on Learning Representations, ICLR 2024, Vienna, Austria, May 7-11, 2024. OpenReview.net

work page 2024
[38]

Yuchen Zhang, Xingan Hao, Feilong Li, Zexi Wang, Dongxiang Li, Mei Li, and Rui Mao. 2025. Unsupervised domain adaptation semantic segmentation method for wheat disease detection based on UAV multispectral images.Comput. Electron. Agric.236 (2025), 110473

work page 2025
[39]

Tao Zhou, Yizhe Zhang, Yi Zhou, Ye Wu, and Chen Gong. 2023. Can SAM Segment Polyps?CoRRabs/2304.07583 (2023)

work page arXiv 2023

[1] [1]

Shubhra Aich and Ian Stavness. 2017. Leaf Counting with Deep Convolutional and Deconvolutional Networks. In2017 IEEE International Conference on Computer Vision Workshops, ICCV Workshops 2017, Venice, Italy, October 22-29, 2017. IEEE Computer Society, 2080–2089

work page 2017

[2] [2]

Mathilde Caron, Hugo Touvron, Ishan Misra, Hervé Jégou, Julien Mairal, Piotr Bojanowski, and Armand Joulin. 2021. Emerging Properties in Self-Supervised Vision Transformers. In2021 IEEE/CVF International Conference on Computer Vision, ICCV 2021, Montreal, QC, Canada, October 10-17, 2021. IEEE, 9630–9640

work page 2021

[3] [3]

Shoufa Chen, Chongjian Ge, Zhan Tong, Jiangliu Wang, Yibing Song, Jue Wang, and Ping Luo. 2022. Adaptformer: Adapting vision transformers for scalable visual recognition.Advances in Neural Information Processing Systems35 (2022), 16664–16678

work page 2022

[4] [4]

Tianrun Chen, Lanyun Zhu, Chaotao Ding, Runlong Cao, Yan Wang, Shangzhan Zhang, Zejian Li, Lingyun Sun, Ying Zang, and Papa Mao. 2023. SAM-Adapter: Adapting Segment Anything in Underperformed Scenes. InIEEE/CVF Interna- tional Conference on Computer Vision, ICCV 2023 - Workshops, Paris, France, October 2-6, 2023. IEEE, 3359–3367

work page 2023

[5] [5]

A ConvNet for the 2020s

Bowen Cheng, Ishan Misra, Alexander G. Schwing, Alexander Kirillov, and Ro- hit Girdhar. 2022. Masked-attention Mask Transformer for Universal Image Segmentation. InIEEE/CVF Conference on Computer Vision and Pattern Recog- nition, CVPR 2022, New Orleans, LA, USA, June 18-24, 2022. IEEE, 1280–1289. doi:10.1109/CVPR52688.2022.00135

work page doi:10.1109/cvpr52688.2022.00135 2022

[6] [6]

Schwing, and Alexander Kirillov

Bowen Cheng, Alexander G. Schwing, and Alexander Kirillov. 2021. Per-Pixel Classification is Not All You Need for Semantic Segmentation. InAdvances in Neu- ral Information Processing Systems 34: Annual Conference on Neural Information Processing Systems 2021, NeurIPS 2021, December 6-14, 2021, virtual, Marc’Aurelio Ranzato, Alina Beygelzimer, Yann N. Daup...

work page 2021

[7] [7]

Etienne David, Simon Madec, Pouria Sadeghi-Tehran, Helge Aasen, Bangyou Zheng, Shouyang Liu, Norbert Kirchgessner, Goro Ishikawa, Koichi Nagasawa, Minhajul A Badhon, et al. 2020. Global wheat head detection (GWHD) dataset: A large and diverse dataset of high-resolution RGB-labelled images to develop and benchmark wheat head detection methods.Plant Phenomics(2020)

work page 2020

[8] [8]

Zehua Fan, Cuiping Liu, Fengyuan Yu, Yongliang Lai, Qiang Wang, Zhiyong Zhang, Juanjuan Zhang, Jinpeng Cheng, Xinming Ma, Xiaohe Gu, and Shuping Xiong. 2026. Dynamic prediction of carbon and nitrogen accumulation in winter wheat grain: Source-sink theory integrated with UAV multispectral imagery. Comput. Electron. Agric.240 (2026), 111228

work page 2026

[9] [9]

Mathieu Pagé Fortin. 2023. Class-Incremental Learning of Plant and Disease Detection: Growing Branches with Knowledge Distillation. InIEEE/CVF Inter- national Conference on Computer Vision, ICCV 2023 - Workshops, Paris, France, October 2-6, 2023. IEEE, 593–603

work page 2023

[10] [10]

Meng-Hao Guo, Cheng-Ze Lu, Qibin Hou, Zhengning Liu, Ming-Ming Cheng, and Shi-Min Hu. 2022. SegNeXt: Rethinking Convolutional Attention De- sign for Semantic Segmentation. InAdvances in Neural Information Process- ing Systems 35: Annual Conference on Neural Information Processing Sys- tems 2022, NeurIPS 2022, New Orleans, LA, USA, November 28 - Decem- ber...

work page 2022

[11] [11]

Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Liang Wang, Weizhu Chen, et al. 2022. Lora: Low-rank adaptation of large language models.Iclr1, 2 (2022), 3

work page 2022

[12] [12]

Jie Jiang, Zhaopeng Fu, Jiayi Zhang, Jinpeng Yang, Wanping Fang, Qiang Cao, Yongchao Tian, Yan Zhu, Weixing Cao, and Xiaojun Liu. 2026. Developing a nitrogen management strategy for winter wheat to enhance economic profit and energy savings using satellite-based management zone mapping at the county scale.Comput. Electron. Agric.244 (2026), 111527

work page 2026

[13] [13]

Lei Ke, Mingqiao Ye, Martin Danelljan, Yifan Liu, Yu-Wing Tai, Chi-Keung Tang, and Fisher Yu. 2023. Segment Anything in High Quality. InAdvances in Neural Information Processing Systems 36: Annual Conference on Neural Information Processing Systems 2023, NeurIPS 2023, New Orleans, LA, USA, December 10 - 16, 2023, Alice Oh, Tristan Naumann, Amir Globerson,...

work page 2023

[14] [14]

Kingma and Jimmy Ba

Diederik P. Kingma and Jimmy Ba. 2015. Adam: A Method for Stochastic Opti- mization. In3rd International Conference on Learning Representations, ICLR 2015, San Diego, CA, USA, May 7-9, 2015, Conference Track Proceedings, Yoshua Bengio and Yann LeCun (Eds.)

work page 2015

[15] [15]

Berg, Wan-Yen Lo, Piotr Dollár, and Ross B

Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloé Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alexander C. Berg, Wan-Yen Lo, Piotr Dollár, and Ross B. Girshick. 2023. Segment Anything. InIEEE/CVF International Conference on Computer Vision, ICCV 2023, Paris, France, October 1-6, 2023. IEEE, 3992–4003

work page 2023

[16] [16]

Yu Lei, Chen Yang, Jinling Zhao, Xue Yang, Chunchun Li, Wenjiang Huang, Hongbo Qiao, and Linsheng Huang. 2026. PPBM-YOLO: A lightweight and accurate detection model for multi-species airborne spores in wheat disease Cross-Infection scenarios.Comput. Electron. Agric.241 (2026), 111325

work page 2026

[17] [17]

Jiachen Li, Jitesh Jain, and Humphrey Shi. 2024. Matting Anything. InIEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2024 - Workshops, Seattle, W A, USA, June 17-18, 2024. IEEE, 1775–1785

work page 2024

[18] [18]

Yang Liu, Muzhi Zhu, Hengtao Li, Hao Chen, Xinlong Wang, and Chunhua Shen

work page

[19] [19]

InThe Twelfth International Conference on Learning Representations, ICLR 2024, Vienna, Austria, May 7-11, 2024

Matcher: Segment Anything with One Shot Using All-Purpose Feature Matching. InThe Twelfth International Conference on Learning Representations, ICLR 2024, Vienna, Austria, May 7-11, 2024. OpenReview.net

work page 2024

[20] [20]

Maxime Oquab, Timothée Darcet, Théo Moutakanni, Huy V. Vo, Marc Szafraniec, Vasil Khalidov, Pierre Fernandez, Daniel Haziza, Francisco Massa, Alaaeldin El-Nouby, Mido Assran, Nicolas Ballas, Wojciech Galuba, Russell Howes, Po- Yao Huang, Shang-Wen Li, Ishan Misra, Michael Rabbat, Vasu Sharma, Gabriel Synnaeve, Hu Xu, Hervé Jégou, Julien Mairal, Patrick La...

work page 2024

[21] [21]

Huihui Pan, Yuanduo Hong, Weichao Sun, and Yisong Jia. 2023. Deep Dual- Resolution Networks for Real-Time and Accurate Semantic Segmentation of Traffic Scenes.IEEE Trans. Intell. Transp. Syst.24, 3 (2023), 3448–3460. doi:10. 1109/TITS.2022.3228042

work page arXiv 2023

[22] [22]

Yongming Rao, Wenliang Zhao, Guangyi Chen, Yansong Tang, Zheng Zhu, Guan Huang, Jie Zhou, and Jiwen Lu. 2022. DenseCLIP: Language-Guided Dense Prediction with Context-Aware Prompting. InIEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2022, New Orleans, LA, USA, June 18-24,

work page 2022

[23] [23]

Vijai Singh and Ak K Misra. 2017. Detection of plant leaf diseases using image seg- mentation and soft computing techniques.Information processing in Agriculture 4, 1 (2017), 41–49

work page 2017

[24] [24]

Yanpeng Sun, Jiahui Chen, Shan Zhang, Xinyu Zhang, Qiang Chen, Gang Zhang, Errui Ding, Jingdong Wang, and Zechao Li. 2024. VRP-SAM: SAM with Vi- sual Reference Prompt. InIEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2024, Seattle, W A, USA, June 16-22, 2024. IEEE, 23565–23574

work page 2024

[25] [25]

Shijie Wang, Zijian Wang, Yadan Luo, Pengfei Zhang, and Xin Yu. [n. d.]. Parameter-Efficient Wheat Disease Segmentation. InDatabases Theory and Appli- cations - 36th Australasian Database Conference, ADC 2025, Sydney, NSW, Australia and Bali, Indonesia, December 4-6, 2025, Proceedings, Renata Borovica-Gajic, Arijit Khan, Bolong Zheng, Xiaoyang Wang, and J...

work page 2025

[26] [26]

Zijian Wang, Radek Zenkl, Latifa Greche, Benoit De Solan, Lucas Bernigaud Samatan, Safaa Ouahid, Andrea Visioni, Carlos A Robles-Zazueta, Francisco Pinto, Ivan Perez-Olivera, et al . 2025. The global wheat full semantic organ segmentation (GWFSS) dataset.Plant Phenomics7, 3 (2025), 100084

work page 2025

[27] [27]

Chapman, and Zi Huang

Tianqi Wei, Xin Yu, Zhi Chen, Scott C. Chapman, and Zi Huang. 2025. Aug- ment to Segment: Tackling Pixel-Level Imbalance in Wheat Disease and Pest Segmentation. InDatabases Theory and Applications - 36th Australasian Database Conference, ADC 2025, Sydney, NSW, Australia and Bali, Indonesia, December 4-6, 2025, Proceedings (Lecture Notes in Computer Scienc...

work page 2025

[28] [28]

Junde Wu, Rao Fu, Huihui Fang, Yuanpei Liu, Zhaowei Wang, Yanwu Xu, Yueming Jin, and Tal Arbel. 2023. Medical SAM Adapter: Adapting Segment Anything Model for Medical Image Segmentation.CoRRabs/2304.12620 (2023)

work page arXiv 2023

[29] [29]

Álvarez, and Ping Luo

Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, José M. Álvarez, and Ping Luo. 2021. SegFormer: Simple and Efficient Design for Seman- tic Segmentation with Transformers. InAdvances in Neural Information Pro- cessing Systems 34: Annual Conference on Neural Information Processing Sys- tems 2021, NeurIPS 2021, December 6-14, 2021, virtual, Marc’Aurelio...

work page 2021

[30] [30]

Mengde Xu, Zheng Zhang, Fangyun Wei, Han Hu, and Xiang Bai. 2023. SAN: Side Adapter Network for Open-Vocabulary Semantic Segmentation.IEEE Trans. Pattern Anal. Mach. Intell.45, 12 (2023), 15546–15561. doi:10.1109/TPAMI.2023. 3311618

work page doi:10.1109/tpami.2023 2023

[31] [31]

Guoyu Yang, Yuan Wang, Daming Shi, and Yanzhong Wang. 2025. Golden Cudgel Network for Real-Time Semantic Segmentation. InIEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2025, Nashville, TN, USA, June 11-15, 2025. Computer Vision Foundation / IEEE, 25367–25376

work page 2025

[32] [32]

Jinyu Yang, Mingqi Gao, Zhe Li, Shang Gao, Fangjing Wang, and Feng Zheng

work page

[33] [33]

Track Anything: Segment Anything Meets Videos.CoRRabs/2304.11968 (2023)

work page arXiv 2023

[34] [34]

Dongshuo Yin, Yiran Yang, Zhechao Wang, Hongfeng Yu, Kaiwen Wei, and Xian Sun. 2023. 1% vs 100%: Parameter-efficient low rank adapter for dense predictions. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 20116–20126

work page 2023

[35] [35]

Wenxi Yue, Jing Zhang, Kun Hu, Yong Xia, Jiebo Luo, and Zhiyong Wang. 2024. SurgicalSAM: Efficient Class Promptable Surgical Instrument Segmentation. In Thirty-Eighth AAAI Conference on Artificial Intelligence, AAAI 2024, Thirty-Sixth Conference acronym ’XX, June 03–05, 2018, Woodstock, NY Wang et al. Conference on Innovative Applications of Artificial In...

work page 2024

[36] [36]

Radek Zenkl, Radu Timofte, Norbert Kirchgessner, Lukas Roth, Andreas Hund, Luc Van Gool, Achim Walter, and Helge Aasen. 2022. Outdoor plant segmentation with deep learning for high-throughput field phenotyping on a diverse wheat dataset.Frontiers in plant science12 (2022), 774068

work page 2022

[37] [37]

Renrui Zhang, Zhengkai Jiang, Ziyu Guo, Shilin Yan, Junting Pan, Hao Dong, Yu Qiao, Peng Gao, and Hongsheng Li. 2024. Personalize Segment Anything Model with One Shot. InThe Twelfth International Conference on Learning Representations, ICLR 2024, Vienna, Austria, May 7-11, 2024. OpenReview.net

work page 2024

[38] [38]

Yuchen Zhang, Xingan Hao, Feilong Li, Zexi Wang, Dongxiang Li, Mei Li, and Rui Mao. 2025. Unsupervised domain adaptation semantic segmentation method for wheat disease detection based on UAV multispectral images.Comput. Electron. Agric.236 (2025), 110473

work page 2025

[39] [39]

Tao Zhou, Yizhe Zhang, Yi Zhou, Ye Wu, and Chen Gong. 2023. Can SAM Segment Polyps?CoRRabs/2304.07583 (2023)

work page arXiv 2023