Fruit Recognition Based on YOLOX*

DOI

Description

Pattern recognition is an urgent problem to be solved in the field of computer vision. In this paper, the network of fruit recognition based on YOLOX is studied. Due to the problem of slow training speed and low accuracy in the classical algorithms, the de-coupling detection head is optimized in YOLOX to overcome the above shortcomings. In terms of data enhancement, a new method combining Mosaic and MixUp is proposed. Through experimental verification, the method proposed in this paper has a great improvement over related algorithms such as YOLOv5, the accuracy is 98.6%, which is increased 5.2%.

Journal

Keywords

Details 詳細情報について

  • CRID
    1390291767548310400
  • DOI
    10.5954/icarob.2022.os11-3
  • ISSN
    21887829
  • Text Lang
    en
  • Data Source
    • JaLC
    • Crossref
  • Abstract License Flag
    Disallowed

Report a problem

Back to top