Electronics Optics & Control, Volume. 30, Issue 5, 44(2023)

Few-Shot Object Detection Algorithm Based on Improved Faster RCNN

DU Yunyan1,2,3, YANG Jinhui1,2,3, LI Hong1,2,3, MAO Yao1,2,3, and JIANG Yu1,2,3
Author Affiliations
  • 1[in Chinese]
  • 2[in Chinese]
  • 3[in Chinese]
  • show less

    At present,most object detection rely on large-scale annotation datasets to ensure the accuracy of detection.In the actual scene,it is very difficult to obtain a large amount of data,and it also takes a lot of manpower and material resources to annotate the data.To solve the problem,a Few-Shot Object Detection algorithm based on Faster RCNN,CA-FSOD,is proposed,which detects the object samples when there are only a few annotated samples in the object category.In order to improve the detection performance,a CBAM-Attention-RPN module is proposed to reduce the number of irrelevant candidate regions.Secondly,a global-local relation detector module is proposed to obtain candidate regions that are more related to the object category by associating the features of a small number of annotated samples and samples to be detected;Finally,a classifier based on cosine Softmax loss is proposed as the classification branch of object detection, which can effectively aggregate the same category features,reduce the intra-class variance and improve the detection accuracy.In order to verify the proposed algorithm,it is trained and tested on MS COCO dataset.The experimental results show that the AP50 is 21.9% for this method,which is better than that of some existing few-shot object detection algorithms.

    Tools

    Get Citation

    Copy Citation Text

    DU Yunyan, YANG Jinhui, LI Hong, MAO Yao, JIANG Yu. Few-Shot Object Detection Algorithm Based on Improved Faster RCNN[J]. Electronics Optics & Control, 2023, 30(5): 44

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category:

    Received: Mar. 19, 2022

    Accepted: --

    Published Online: Nov. 29, 2023

    The Author Email:

    DOI:10.3969/j.issn.1671-637x.2023.05.009

    Topics