PARTIAL OCCLUSION OBJECT DETECTION BASED ON IMPROVED MASK-RCNN
DOI:
https://doi.org/10.15282/ijsecs.10.1.2024.2.0120Keywords:
Vision detection, Mask-RCNN, Recognition, Accuracy, SpeedAbstract
In the grasping task of industrial robots, multi-target objects are often placed in disorder or even partially occlusion or stacked, which brings certain difficulties to visual detection such as accuracy and real-time. The traditional Mask-RCNN algorithm can achieve high detection accuracy in scene which the target objects are neatly placed, but in the complex scenarios such as disorderly placed or partially occlusion is still have space for improvement in accuracy and speed. Mask-RCNN introduces the mask head structure to achieve pixel level segmentation mask prediction, it achieves high detection accuracy but increases the amount of calculation, this cause the detection speed is limited. To deal the above problems, this paper proposes an indirect frame subtraction for loss function to improved Mask-RCNN, which uses adjacent frames as comparison templates to find differences for image, that is, after one recognition, the previous recognition result is used as the background, and the next change is used as the target. thereby improving the recognition accuracy, reducing the repetitive estimation of regions, and improving the detection accuracy and running speed. Through experiments on self-made datasets, it is proved that new method can improve the image recognition accuracy about 2.3%, another, image recognition time has been reduced by 9% and the FPS value is improved by 6 frames which indicate the speed was improved. The research has important reference significance for the realization of robot flexible grasping task in intelligent manufacturing environment.
References
Fu Donghao, “Based on visual target recognition with seven degrees of freedom robot grasping study,” Shenyang university of technology, 2023. doi: 10.27322 /, dc nki. Gsgyu. 2023.000388.
WEI.H,PAN.S,MA.G,et al, “Vision-guided hand-eye coordination for robotic grasping and its application in tangram puzzles,”AIVolume 2, Issue 2. PP 209-228, 2021,doi:10.3390/AI2020013.
Yun.Ke Chen et al,“K-TIG welding seam tracking based on Mask-RCNN model Systematic research,”Master's thesis of South China University of Technology,pp.9-12,2021,doi:10561TG409.
Zhu.Jiahui,“Research on cross-weather road scene re-recognition algorithm,” Beijing University of Posts and Telecommunications, pp.35-36,2018,doi:TP391.41.
Ma Hao,“Research on crowd counting algorithm based on Convolutional Neural Network,” University of Science and Technology of China, pp.15,2018,doi:10.27791/d.cnki.ghegy.2023.000360.
Girshick,Ross B.et al,“Rich Feature Hierarchies for Accurate Object Detection and Semantic Segm entati on,” IEEE Conference on Computer Vision and Pattern Recognition（2014）,pp.580-587,2014,doi:10.1109/CVPR.2008.4587671.
Salti.S, Tombari.F, Di.Stefano L. SHOT, “Unique sign atures of histograms for surface and texture description,” Computer Vision and Image Understan ding, 125251-264, 2014,doi:10.1007/978-3-642-15558-1_26.
Xiang.Y, Schmidt.T, Narayan.an V, et al,“PoseCNN: A Convolutional Neural Network for 6D Object Pose Estimation in Cluttered Scenes,” 14th Conference on Robotics -Science and Sy stems, pp.123-136,2018,doi:10.15607/rss.2018.xiv.019.
Tejani A, Tang D H, Kouskouridas R, et al, “Latent-Class Hough Forests for 3D Object Detection and Pose Estimation,” 13th European Conferen ce on Computer Vision(ECCV),pp.462-477,2014 .doi:10.1109/TPAMI.2017.2665623.
LENZ.I,LEE.H,SAXENA.A,“Deep learning for detecting robotic grasps,”The International Journal of Robotics Research,34(4),pp.705–724.2015,doi:10.1177/0278364914549607.
Li Longyan,“Based on visual intelligent fetching of mechanical arm technology research,” North China university of water conservancy and hydropower, 2023. doi: 10.27144 /, dc nki. GHBSC. 2023.000205.
Kang.Wei H, Yan.Jun.L,“Mask-RCNN Segmentation Based on 3D Points Cloud Matching for Fish Dimension Measurement,”39th Chinese Control Conference . doi: 10.26914 / Arthur c. nkihy. 2020.038310.
Yan.Na L, DanYang D,“Design and research of bridge crack detection method based on Mask-RCNN,”Applied Optics 43.01(2022):100-105+118. doi: CNKI: SUN: YYGX. 0.2022-01-016.
Da.Song L, “Automatic crack detection of aircraft structure based on improved Mask‐RCNN method,”Journal of Vibration Measurement & Diagnosis,2021, (03),pp.487-494,2021,doi:10.16450/j.cnki.issn.1004-6801.2021.03.010.
Zeng.A, Song.S, Yu.K,“Robotic Pick-and-Place of Novel Objects in Clutter with Multi-Affordance Grasping and Cross-Domain Image Matching,” 2018 EEE International Conference on Robotics and Autom action (ICRA),pp.3750-3757,2018,doi:10.1109/ICRA.2018.8461044.
Jiang.Chuan.Y,“Pavement crack detection based improved Mask-RCNN,”Television Technology 46.06 (2022):7-9+19. doi: 10.16280 / j.v ideoe 2022.06.003.
Yang.Hua.T, Hong.Jian.W,“Feature Extraction for Side Scan Sonar Image Based Deep Learning,” Harbin engineering university, 2022. doi: 10.27060 /, dc nki. Ghbcu. 2022.002320.
Bo.Sen F, Yun Bo Z, “Road information detection algorithm based on improved Mask-RCNN,”Journal of Information Science and Technology University of Beijing (Natural Science Edition). 37(03),pp.19-23,2022,doi:10.11772/j.issn.1001-9081.2020030357.
Qi.C, Rui.Cheng L,“Power line identification method based on improved Mask-RCNN,”Journal of Shantou University (Natural Science Edition). 37(02),pp.65-68,2022,doi:1001 - 4217（2022）02 - 0043 - 07.
Bin.W,“Research on robot capture detection algorithms based depth image and deep learning,”Zhejiang University(2019).
Na.L,Tao.H,“Building extraction from Mask-RCNN high-resolution remote sensing images,” Remote Sensing Information (03),1-6. doi:CNKI:SUN:YGXX.0.2022-03-001.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2024 QINGCHUAN LIU, Azmi Ayub Muhammad , Fazlina Ahmat Ruslan, Mohd Nor Azmi Ab Patar, Shuzlina Abdul-Rahman
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
