目标检测中的平均精度均值(mean average precision,mAP)是评价检测性能的重要指标。本文采用AP50和AP75作为评价指标,分别代表IoU阈值为0.5和0.75时各个类别AP的均值。AP计算的是精确度(precision)和召回率(recall)曲线下的面积,其值越大表示模型检测精度越高,精确度、召回率和AP的计算公式如式(6)~(8)所示。
PurkaitP, ZhaoC, ZachC.SPP-Net:deep absolute pose regression with synthetic views[C]//CVPR.Salt Lake City:IEEE,2018:1101-1256.
[4]
GirshickR.Fast R-CNN[C]//2015 IEEE International Conference on Computer Vision (ICCV).Santiago:IEEE,2015:1440-1448.
[5]
LinT Y, GoyalP, GirshickR,et al.Focal loss for dense object detection[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2020,42(2):318-327.
KisantalM, WojnaZ, MurawskiJ,et al.Augmentation for small object detection[EB/OL].(2019-02-19)[2023-01-23].
[9]
LinT Y, DollárP, GirshickR,et al.Feature pyramid networks for object detection[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).Honolulu:IEEE,2017:936-944.
[10]
LiuS, QiL, QinH F,et al.Path aggregation network for instance segmentation[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Salt Lake City:IEEE,2018:8759-8768.
[11]
YangC, HuangZ H, WangN Y.QueryDet:cascaded sparse query for accelerating high-resolution small object detection[C]//2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).New Orleans:IEEE,2022:13658-13667.
[12]
PrajitR, NikiP, AshishV,et al.Stand-alone self-attention in vision models[C]//Proceedings of the 33rd International Conference on Neural Information Processing Systems.Red Hook:IEEE.2019:68-80.
LiuZ, LinY T, CaoY,et al.Swin Transformer:hierarchical Vision Transformer using Shifted Windows[C]//2021 IEEE/CVF International Conference on Computer Vision (ICCV).Montreal:IEEE,2021:9992-10002.
[15]
GevorgyanZ.SIoU loss:more powerful learning for bounding box regression[EB/OL].(2022-05-25)[2024-04-23].
[16]
DuD W, ZhuP F, WenL Y,et al.VisDrone-DET2019:the vision meets drone object detection in image challenge results[C]//2019 IEEE/CVF International Conference on Computer Vision Workshop (ICCVW).Seoul:IEEE,2019:213-226.
[17]
LiZ, PengC, YuG,et al.Light-head R-CNN:indefense of two-stage object detector[EB/OL].(2018-06-04)[2024-04-23].
[18]
CaiZ W, VasconcelosN.Cascade R-CNN:delving into high quality object detection[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Salt Lake City:IEEE,2018:6154-6162.
[19]
RenS Q, HeK M, GirshickR,et al.Faster R-CNN:towards real-time object detection with region proposal networks[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2017,39(6):1137-1149.
[20]
LawH, DengJ.CornerNet:detecting objects as paired keypoints[J].International Journal of Computer Vision,2020,128(3):642-656.
[21]
GeZ, LiuS T, WangF,et al.YOLOX:exceeding YOLO series in 2021[EB/OL].(2021-07-18)[2024-04-23].
[22]
ZhuX K, LyuS C, WangX,et al.TPH-YOLOv5:improved YOLOv5 based on transformer prediction head for object detection on drone-captured scenarios[C]//2021 IEEE/CVF International Conference on Computer Vision Workshops (ICCVW).Montreal:IEEE,2021:2778-2788.