基于遮挡增强知识蒸馏的果实识别

史源平; 马彦恒; 耿亮; 褚丽娜; 李伟; 李秉璇

doi:null

您当前的位置：

首页 >

文章列表页 >

基于遮挡增强知识蒸馏的果实识别

光学测量与感知 | 更新时间：2025-09-22

- 基于遮挡增强知识蒸馏的果实识别
- Fruit recognition based on occlusion enhanced knowledge distillation
- 光学技术 2025年51卷第5期页码：617-625
- 作者机构：
  
  1.陆军工程大学石家庄校区，河北石家庄 050003
  2.石家庄学院机电工程学院，河北石家庄 050035
  3.石家庄市农业机器人智能感知重点实验室，河北石家庄 050035
- 作者简介：
  
  史源平（1970— ），女，教授，博士，从事自动检测、目标识别方面的研究。
  马彦恒（1968— ），男，教授，博士，从事雷达信号处理方面的研究。1102061@sjzc.edu.cn
- 基金信息：
  
  河北省科技厅重点研发计划(21327213D)
- DOI：
  中图分类号： TP391
- 收稿：2025-06-17，
  
  修回：2025-07-06，
  
  纸质出版：2025-09-20
- 稿件说明：
移动端阅览
史源平,马彦恒,耿亮等.基于遮挡增强知识蒸馏的果实识别[J].光学技术,2025,51(05):617-625.

SHI Yuanping,MA Yanheng,GENG Liang,et al.Fruit recognition based on occlusion enhanced knowledge distillation[J].OPTICAL TECHNIQUE,2025,51(05):617-625.
史源平,马彦恒,耿亮等.基于遮挡增强知识蒸馏的果实识别[J].光学技术,2025,51(05):617-625. DOI：

SHI Yuanping,MA Yanheng,GENG Liang,et al.Fruit recognition based on occlusion enhanced knowledge distillation[J].OPTICAL TECHNIQUE,2025,51(05):617-625. DOI：

摘要

果实被枝叶遮挡时会显著削弱采摘机器人视觉系统的识别精度。针对这一问题，文章提出一种基于“遮挡增强知识蒸馏”的果实识别框架。该框架在构建的遮挡数据集上，通过引入遮挡信息约束多尺度语义特征对齐，并结合指数移动平均策略稳定训练过程。在MinneApple数据集上的实验结果表明，AP@[0.5:0.05:0.95]、AP@0.5、AP@0.75、小目标AP与大目标AP分别达到0.744、0.946、0.793、0.674和0.976，较主流检测模型平均提升0.178。研究证明，方法能够在复杂遮挡场景下显著提高采摘机器人的视觉感知精度，为智慧农业果实自动采摘提供有效技术支撑。

Abstract

In order to solve the problem that the fruit covered by branches and leaves will affect the optical recognition effect of the picking robot

a recognition method of "shading enhanced distillation" is proposed. On the self-made occlusion data set

the method uses occlusion information to standardize the learning of semantic alignment features

and uses exponential moving average to enhance the training stability. Experiments are carried out on the MinneApple dataset. The AP@[0.5:0.05:0.95] is 0.744. The AP@0.5 is 0.946. The AP@0.75 is 0.793. The AP for small objects is 0.674. The AP for large objects is 0.976. These results show an average improvement of 0.178 over mainstream detection models. The experimental results show that this method can provide effective technical support for the accurate visual perception of picking robot in occluded scenes.

关键词

Keywords

references

Moisiadis V , Tsolakis N , Katikaridis D , et al . Mobile robotics in agricultural operations: a narrative review on planning aspects [J ] . Applied Sciences , 2020 , 10 （ 10 ）: 3453 .

Wang Quanyu , He Jin , Lu Caiyun , et al . Modelling and control methods in path tracking control for autonomous agricultural vehicles: a review of state of the art and challenges [J ] . Applied Sciences , 2023 , 13 （ 12 ）: 7155 .

Aslan M F , Durdu A , Sabanci K , et al . A comprehensive survey of the recent studies with UAV for precision agriculture in open fields and greenhouses [J ] . Applied Sciences , 2022 , 12 （ 3 ）: 1047 .

Lai Yuhao , Ma Ruijun , Chen Yu , et al . A pineapple target detection method in a field environment based on improved YOLOv7 [J ] . Applied Sciences , 2023 , 13 （ 4 ）: 2691 .

Girshick R . Fast R-CNN [C ] // 2015 IEEE International Conference on Computer Vision （ICCV） . Santiago, USA : IEEE , 2015 : 1440 — 1448 .

Ren S , He K , Girshick R , et al . Faster R-CNN: towards real-time object detection with region proposal networks [J ] . IEEE Transactions on Pattern Analysis & Machine Intelligence , 2017 , 39 （ 6 ）: 1137 — 1149 .

Lucena F , Breunig FM , Kux H . The combined use of UAV-based RGB and DEM images for the detection and delineation of orange tree crowns with mask R-CNN: an approach of labeling and unified framework [J ] . Future Internet , 2022 , 14 （ 10 ）: 275 .

Redmon J , Divvala S , Girshick R , et al . You only look once: unified, real-time object detection [C ] // 2016 IEEE Conference on Computer Vision and Pattern Recognition （CVPR） . Las Vegas, USA : IEEE , 2016 : 779 — 788 .

Liu Wei , Anguelov D , Szegedy C , et al . SSD: single shot multibox detector [C ] // Computer Vision-ECCV 2016 . Amsterdam, The Netherlands : Springer , 2016 : 21 — 37 .

Zhu X , Su W , Lu L , et al . Deformable DETR: deformable transformers for end-to-end object detection [C ] // International Conference on Learning Representations （ICLR） . Vienna, Austria （Virtual） : ICLR , 2021 .

Zhang H , Li F , Liu S , et al . DINO: dETR with improved denoising anchor boxes for end-to-end object detection [C ] // International Conference on Learning Representations （ICLR） . Kigali, Rwanda : ICLR , 2023 .

Liu S , Zeng Z , Ren T , et al . Grounding DINO: marrying DINO with grounded pre-training for open-set object detection [C ] // Computer Vision – ECCV 2024 . Cham : Springer , 2024 : 38 — 55 .

Kirillov A , Mintun E , Ravi N , et al . Segment anything [C ] // 2023 IEEE/CVF International Conference on Computer Vision （ICCV） . Paris, France : Ithaca: Cornell University Library , 2023 : 3992 — 003 .

Guo J , Han K , Wang Y , et al . Distilling object detectors via decoupled features [C ] // Proceedings （IEEE Computer Society Conference on Computer Vision and Pattern Recognition Online） . IEEE , 2021 : 2154 — 64 .

Chang J , Wang S , Xu HM , et al . Detrdistill: a universal knowledge distillation framework for detr-families [C ] // IEEE/CVF International Conference on Computer Vision , ICCV 2023 . Paris,France : IEEE ,2023: 6898 — 6908 .

Carion N , Massa F , Synnaeve G , et al . End-to-end object detection with transformers [C ] // Computer Vision-ECCV 2020 . ECCV 2020 . Online : Springer ,2020: 0213 — 229 .

Saleh K , Sz´en´asi S , V´amossy Z . Occlusion handling in generic object detection: a review [C ] // 2021 IEEE 19th World Symposium on Applied Machine Intelligence and Informatics （SAMI） . Online : IEEE , 2021 : 000477 — 000484 .

Ruan J , Cui H , Huang Y , et al . A review of occluded objects detection in real complex scenarios for autonomous driving [J ] . Green Energy and Intelligent Transportation , 2023 , 2 （ 3 ）: 100092 .

Touvron H , Cord M , Douze M , et al . Training data-efficient image transformers & distillation through attention [C ] // International Conference on Machine Learning （Online）:PMLR , 2021 : 10347 — 10357 .

Yang Z , Li Z , Shao M , et al . Masked generative distillation [C ] // Computer Vision-ECCV 2022 . ECCV 2022 . Tel Aviv, Israel : Springer ,2022: 53 — 69 .

Chen Q , Chen X , Wang J , et al . Group DETR: fast DETR training with group-wise one-to-many assignment [C ] // Proceedings of the IEEE/CVF International Conference on Computer Vision （ICCV） . Paris, France : IEEE/CVF , 2023 : 6633 — 6642 .

Chen K , Wang J , Pang J , et al . MMDetection: open mmlab detection toolbox and benchmark [J ] . arXiv preprint arXiv: 1906 ,07155.2019.

Gao Z , Wang L , Han B , et al . Adamixer: a fast-converging query-based object detector [C ] // 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition （CVPR） . Louisiana, USA : IEEE , 2022 : 5364 — 5373 .

Chen S , Cheng T , Fang J , et al . TinyDet: accurately detecting small objects within 1 GFLOPs [J ] . Science China Information Sciences , 2023 , 66 : 119102 .

Dosovitskiy A , Beyer L , Kolesnikov A , et al . An image is worth 16×16 words: transformers for image recognition at scale [C ] // International Conference on Learning Representations （ICLR） . Virtual Event, Austria : ICLR , 2021 .

He K , Fan H , Wu Y , et al . Momentum contrast for unsupervised visual representation learning [C ] // 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition （CVPR） . State of Washington, USA : IEEE , 2020 : 9729 — 9738 .

Grill J B , Strub F , Altch´e F , et al . Bootstrap your own latent a new approach to self-supervised learning [J ] . Advances in Neural Information Processing Systems , 2020 , 33 : 21271 — 21284 .

Hani N , Roy P , Isler V . MinneApple: a benchmark dataset for apple detection and segmentation .[J ] . IEEE Robotics and Automation Letters , 2020 , 5 （ 2 ）: 852 — 858 .

Kingma D P , Ba J . Adam: a method for stochastic optimization [C ] // International Conference on Learning Representations （ICLR） . San Diego,USA : ICLR , 2015 .

Sung-Pil S , Sang-Yum L . Feasibility of efficientdet-d3 for accurate and efficient void detection in GPR images [J ] . Infrastructures , 2025 , 10 （ 6 ）: 140 .

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于平面激光散射的人体边界层及口呼吸可视化研究