A D-G-YOLOV3 algorithm was proposed to identify and judge recyclables, which introduced a dense
feature network to replace the feature pyramid network. The network closely connects and fits the feature maps
and simulates human judgment mechanism. A three-stage judgment is made for judgment objects with lower
confidence. Based on the judgment of the original image, the second-stage judgment is carried out after the
channel contrast is increased. Finally, sampling is performed on the region of interest where the second-stage
confidence score wins for the third stage of judgment, and then judgment result is sent to the gated recurrent
unit network for final inference. The result shows that through experiments on the same recyclables data set, the
algorithm reduces the missed detection rate by 15.54%, and the false detection rate by 0.97%, while improves the
accuracy rate by 16.51%.
SHI Jichao1 (石继超), WANG Ziheng2 (王子恒), ZHAO Xianchao1 (赵现朝), ZHANG Zhinan1∗ (张执南)
. Target Detection Algorithm Based On Human Judge Mechanism[J]. Journal of Shanghai Jiaotong University(Science), 2022
, 27(5)
: 660
-670
.
DOI: 10.1007/s12204-022-2450-5
[1] ZHAO Y Q, RAO Y, DONG S P, et al. Survey on deep learning object detection [J]. Journal of Image and Graphics, 2020, 25(4): 629-654 (in Chinese)
[2] GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation [C]//2014 IEEE Conference on Computer Vision and Pattern Recognition. Columbus, OH, USA: IEEE, 2014: 580-587.
[3] REN S Q, HE K M, GIRSHICK R, et al. Faster RCNN: Towards real-time object detection with region proposal networks [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6): 1137- 1149.
[4] LIU W, ANGUELOV D, ERHAN D, et al. SSD: single shot MultiBox detector [M]//Computer vision-ECCV 2016. Cham: Springer, 2016.
[5] LIN T Y, GOYAL P, GIRSHICK R, et al. Focal loss for dense object detection [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2020, 42(2): 318-327.
[6] REDMON J, FARHADI A. YOLOv3: an incremental improvement [EB/OL]. (2018-04-08). https://arxiv.org/abs/1804.02767.
