{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,11]],"date-time":"2026-08-11T06:27:58Z","timestamp":1786429678172,"version":"build-2736575974"},"reference-count":42,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,8,11]],"date-time":"2026-08-11T00:00:00Z","timestamp":1786406400000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Plant Sci."],"abstract":"<jats:p>\n                    Automated litchi harvesting demands clustered fruit segmentation with both high accuracy and real-time performance, yet high computational complexity and parameter redundancy pose critical bottlenecks for practical deployment. To tackle these challenges, we propose a lightweight DSD-UNet method for fast and high-precision segmentation of litchi clusters. The DSD-UNet incorporates depthwise separable convolutions to replace traditional convolutional structures, significantly reducing parameters and computational complexity. To further enhance the model\u2019s performance, we introduce a dual cross-attention mechanism into the skip connections,which integrates spatial and channel attention to adaptively strengthen the representation of target regions in feature maps and improve sensitivity to slender stem features. Additionally, we develop a composite loss function (BL-Loss), combining binary cross-entropy with level-set loss, and optimize it using a staged training strategy to alleviate within-foreground pixel imbalance and preserve the boundary continuity of slender fruit\u2013stem connections. Experimental results under varying illumination conditions demonstrate that DSD-UNet achieves a mean Intersection over Union (\n                    <jats:italic>mIoU<\/jats:italic>\n                    ) of 91.40% and an\n                    <jats:italic>F1-score<\/jats:italic>\n                    of 95.42% for the segmentation of clustered litchi, with an inference speed of 15.41 frames per second and a parameter count of only 5.16M. Compared with the original U-Net, our DSD-UNet model improves\n                    <jats:italic>mIoU<\/jats:italic>\n                    and\n                    <jats:italic>F1-score<\/jats:italic>\n                    by 3.08 and 1.81 percentage points, respectively, increases inference speed by 138.9%, and achieves a 61.5% reduction in parameter count, confirming that the proposed method effectively enhances segmentation accuracy while maintaining model lightweightness. These results underscore the potential of our method to significantly improve automated harvesting processes in litchi production.\n                  <\/jats:p>","DOI":"10.3389\/fpls.2026.1855288","type":"journal-article","created":{"date-parts":[[2026,8,11]],"date-time":"2026-08-11T05:35:54Z","timestamp":1786426554000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["A lightweight DSD-UNet-based method for clustered litchi segmentation"],"prefix":"10.3389","volume":"17","author":[{"given":"Zhiping","family":"Tan","sequence":"first","affiliation":[{"name":"College of Electronics and Information, Guangdong Polytechnic Normal University","place":["Guangzhou, China"]},{"name":"College of Mechanical and Electrical Engineering, Fujian Agriculture and Forestry University","place":["Fuzhou, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yihao","family":"Huang","sequence":"additional","affiliation":[{"name":"College of Electronics and Information, Guangdong Polytechnic Normal University","place":["Guangzhou, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Dapeng","family":"Ye","sequence":"additional","affiliation":[{"name":"College of Mechanical and Electrical Engineering, Fujian Agriculture and Forestry University","place":["Fuzhou, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,8,11]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","DOI":"10.1016\/j.engappai.2023.107139","article-title":"Dual cross-attention for medical image segmentation","volume":"126","author":"Ates","year":"2023","journal-title":"Eng. Appl. Artif. Intell."},{"key":"B2","doi-asserted-by":"publisher","DOI":"10.1016\/j.compag.2023.108362","article-title":"Accurate detection and depth estimation of table grapes and peduncles for robot harvesting, combining monocular depth estimation and CNN methods","volume":"215","author":"Coll-Ribes","year":"2023","journal-title":"Comput. Electron. Agric."},{"key":"B3","article-title":"Improved regularization of convolutional neural networks with Cutout","author":"DeVries","year":"2017"},{"key":"B4","doi-asserted-by":"publisher","first-page":"107940","DOI":"10.2139\/ssrn.4314624","article-title":"Maize crop row recognition algorithm based on improved UNet network","volume":"210","author":"Diao","year":"2023","journal-title":"Comput. Electron. Agric."},{"key":"B5","doi-asserted-by":"publisher","first-page":"103","DOI":"10.1016\/j.compag.2017.05.019","article-title":"An yield estimation in citrus orchards via fruit detection and counting using image processing","volume":"140","author":"Dorj","year":"2017","journal-title":"Comput. Electron. Agric."},{"key":"B6","doi-asserted-by":"publisher","DOI":"10.1016\/j.compag.2023.108056","article-title":"Tomato 3D pose detection algorithm based on keypoint detection and point cloud processing","volume":"212","author":"Du","year":"2023","journal-title":"Comput. Electron. Agric."},{"key":"B7","doi-asserted-by":"publisher","first-page":"64","DOI":"10.1016\/j.aiia.2024.06.004","article-title":"Computer vision in smart agriculture and precision farming: techniques and applications","volume":"13","author":"Ghazal","year":"2024","journal-title":"Artif. Intell. Agric."},{"key":"B8","doi-asserted-by":"publisher","DOI":"10.3390\/s19194091","article-title":"Lychee fruit detection based on monocular machine vision in orchard environment","volume":"19","author":"Guo","year":"2019","journal-title":"Sensors"},{"key":"B9","article-title":"Yolov11: an overview of the key architectural enhancements","author":"Khanam","year":"2024","journal-title":"arXiv preprint arXiv:2410.17725"},{"key":"B10","doi-asserted-by":"publisher","DOI":"10.4258\/hir.2016.22.4.351","article-title":"Book review: Deep learning","volume":"22","author":"Kim","year":"2016","journal-title":"Healthc Inform. Res."},{"key":"B11","doi-asserted-by":"publisher","DOI":"10.1016\/j.asoc.2022.108556","article-title":"Bayberry segmentation in a complex environment based on a multi-module convolutional neural network","volume":"119","author":"Lei","year":"2022","journal-title":"Appl. Soft Comput."},{"key":"B12","doi-asserted-by":"publisher","first-page":"108605","DOI":"10.2139\/ssrn.4511085","article-title":"An efficient detection method for litchi fruits in a natural environment based on improved YOLOv7-Litchi","volume":"217","author":"Li","year":"2024","journal-title":"Comput. Electron. Agric."},{"key":"B13","doi-asserted-by":"publisher","DOI":"10.1016\/j.compag.2024.109748","article-title":"Kiwifruit segmentation and identification of picking point on its stem in orchards","volume":"229","author":"Li","year":"2025","journal-title":"Comput. Electron. Agric."},{"key":"B14","doi-asserted-by":"publisher","DOI":"10.1016\/j.compag.2024.109731","article-title":"A method for segmentation and localization of tomato lateral pruning points in complex environments based on improved YOLOV5","volume":"229","author":"Liang","year":"2025","journal-title":"Comput. Electron. Agric."},{"key":"B15","doi-asserted-by":"publisher","first-page":"17880","DOI":"10.1038\/s41598-023-45186-0","article-title":"Research on anti-clogging of ore conveyor belt with static image based on improved Fast-SCNN and U-Net","volume":"13","author":"Liu","year":"2023","journal-title":"Sci. Rep."},{"key":"B16","doi-asserted-by":"publisher","DOI":"10.1016\/j.compag.2024.109531","article-title":"Y-HRNet: research on multi-category cherry tomato instance segmentation model based on improved YOLOv7 and HRNet fusion","volume":"227","author":"Liu","year":"2024","journal-title":"Comput. Electron. Agric."},{"key":"B17","doi-asserted-by":"publisher","DOI":"10.1016\/j.compag.2023.107834","article-title":"ORB-Livox: a real-time dynamic system for fruit detection and localization","volume":"209","author":"Liu","year":"2023","journal-title":"Comput. Electron. Agric."},{"key":"B18","first-page":"213","article-title":"PIDNet-Lite: a lightweight semantic segmentation model for power line recognition","author":"Liu","year":"2025"},{"key":"B19","doi-asserted-by":"crossref","DOI":"10.1109\/CVPR.2015.7298965","article-title":"Fully convolutional networks for semantic segmentation","author":"Long","year":"2015"},{"key":"B20","doi-asserted-by":"publisher","first-page":"230","DOI":"10.1016\/j.media.2018.06.005","article-title":"Deep learning and conditional random fields-based depth estimation and topographical reconstruction from conventional endoscopy","volume":"48","author":"Mahmood","year":"2018","journal-title":"Med. Image Anal."},{"key":"B21","first-page":"345","article-title":"Litchi image segmentation algorithm based on sparse field level set","volume":"27","author":"Mao","year":"2011","journal-title":"Trans. Chin. Soc Agric. Eng."},{"key":"B22","doi-asserted-by":"publisher","first-page":"107691","DOI":"10.2139\/ssrn.4162665","article-title":"ResDense-focal-DeepLabV3+ enabled litchi branch semantic segmentation for robotic harvesting","volume":"206","author":"Peng","year":"2023","journal-title":"Comput. Electron. Agric."},{"key":"B23","first-page":"234","article-title":"U-Net: convolutional networks for biomedical image segmentation","volume-title":"Medical Image Computing and Computer-Assisted Intervention \u2013 MICCAI 2015","author":"Ronneberger","year":"2015"},{"key":"B24","doi-asserted-by":"publisher","DOI":"10.1016\/j.compag.2020.105933","article-title":"Canopy segmentation and wire reconstruction for kiwifruit robotic harvesting","volume":"181","author":"Song","year":"2021","journal-title":"Comput. Electron. Agric."},{"key":"B25","doi-asserted-by":"publisher","DOI":"10.3389\/fpls.2025.1612642","article-title":"P4CN-YOLOv5s: a passion fruit pests detection method based on lightweight-improved YOLOv5s","volume":"16","author":"Tan","year":"2025","journal-title":"Front. Plant Sci."},{"key":"B26","doi-asserted-by":"publisher","first-page":"176","DOI":"10.25165\/j.ijabe.20251803.8935","article-title":"Citrus fruit detection based on an improved YOLOv5 under natural orchard conditions","volume":"18","author":"Tang","year":"2025","journal-title":"Int. J. Agric. Biol. Eng."},{"key":"B27","first-page":"1451","article-title":"Understanding convolution for semantic segmentation","author":"Wang","year":"2018"},{"key":"B28","doi-asserted-by":"publisher","DOI":"10.1016\/j.patcog.2024.110324","article-title":"Self-supervised local rotation-stable descriptors for 3D ultrasound registration using translation equivariant FCN","volume":"150","author":"Wang","year":"2024","journal-title":"Pattern Recognit."},{"key":"B29","doi-asserted-by":"publisher","DOI":"10.1016\/j.compag.2024.109473","article-title":"Litchi picking points localization in natural environment based on the Litchi-YOSO model and branch morphology reconstruction algorithm","volume":"226","author":"Wang","year":"2024","journal-title":"Comput. Electron. Agric."},{"key":"B30","doi-asserted-by":"publisher","DOI":"10.1016\/j.cmpb.2021.106210","article-title":"Medical image recognition and segmentation of pathological slices of gastric cancer based on Deeplab v3+ neural network","volume":"207","author":"Wang","year":"2021","journal-title":"Comput. Methods Programs Biomed."},{"key":"B31","doi-asserted-by":"publisher","DOI":"10.1016\/j.compag.2024.109551","article-title":"Wolfberry recognition and picking-point localization technology in natural environments based on improved Yolov8n-Pose-LBD","volume":"227","author":"Wang","year":"2024","journal-title":"Comput. Electron. Agric."},{"key":"B32","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1109\/tgrs.2023.3314641","article-title":"CMTFNet: CNN and multiscale transformer fusion network for remote-sensing image semantic segmentation","volume":"61","author":"Wu","year":"2023","journal-title":"IEEE Trans. Geosci. Remote Sens."},{"key":"B33","doi-asserted-by":"publisher","DOI":"10.1016\/j.compag.2025.110564","article-title":"MixSegNext: a CNN-Transformer hybrid model for semantic segmentation and picking point localization algorithm of Sichuan pepper in natural environments","volume":"237","author":"Xiang","year":"2025","journal-title":"Comput. Electron. Agric."},{"key":"B34","first-page":"14046","article-title":"MaSS13K: a matting-level semantic segmentation benchmark","author":"Xie","year":"2025"},{"key":"B35","doi-asserted-by":"publisher","DOI":"10.1016\/j.compag.2023.108217","article-title":"Real-time localization and 3D semantic map reconstruction for unstructured citrus orchards","volume":"213","author":"Xiong","year":"2023","journal-title":"Comput. Electron. Agric."},{"key":"B36","doi-asserted-by":"publisher","DOI":"10.1016\/j.compmedimag.2021.102026","article-title":"Unified focal loss: generalising Dice and cross entropy-based losses to handle class imbalanced medical image segmentation","volume":"95","author":"Yeung","year":"2022","journal-title":"Comput. Med. Imaging Graphics"},{"key":"B37","doi-asserted-by":"publisher","DOI":"10.3389\/fbioe.2022.861286","article-title":"Real-time target detection method based on lightweight convolutional neural network","volume":"10","author":"Yun","year":"2022","journal-title":"Front. Bioeng. Biotechnol."},{"key":"B38","doi-asserted-by":"publisher","first-page":"147","DOI":"10.1016\/j.procs.2023.08.152","article-title":"A dual-path multi-scale feature fusion decoder for SegFormer","volume":"222","author":"Zhang","year":"2023","journal-title":"Proc. Comput. Sci."},{"key":"B39","doi-asserted-by":"publisher","DOI":"10.1016\/j.compag.2021.106398","article-title":"A method for litchi picking points calculation in natural environment based on main fruit bearing branch detection","volume":"189","author":"Zhong","year":"2021","journal-title":"Comput. Electron. Agric."},{"key":"B40","doi-asserted-by":"publisher","DOI":"10.1016\/j.cmpb.2020.105897","article-title":"Coronary angiography image segmentation based on PSPNet","volume":"200","author":"Zhu","year":"2021","journal-title":"Comput. Methods Programs Biomed."},{"key":"B41","doi-asserted-by":"publisher","DOI":"10.3390\/s19132910","article-title":"Assessment of external properties for identifying banana fruit maturity stages using optical imaging techniques","volume":"19","author":"Zhuang","year":"2019","journal-title":"Sensors"},{"key":"B42","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1016\/j.biosystemseng.2019.08.016","article-title":"Computer vision-based localisation of picking points for automatic litchi harvesting applications towards natural scenarios","volume":"187","author":"Zhuang","year":"2019","journal-title":"Biosyst. Eng."}],"container-title":["Frontiers in Plant Science"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fpls.2026.1855288\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,8,11]],"date-time":"2026-08-11T05:35:56Z","timestamp":1786426556000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fpls.2026.1855288\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,8,11]]},"references-count":42,"alternative-id":["10.3389\/fpls.2026.1855288"],"URL":"https:\/\/doi.org\/10.3389\/fpls.2026.1855288","relation":{},"ISSN":["1664-462X"],"issn-type":[{"value":"1664-462X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,8,11]]},"article-number":"1855288"}}