改进的Cascade RCNN行人检测算法研究

doi:10.3778/j.issn.1002-8331.2009-0298

摘要/Abstract

摘要： 在复杂路况下的行人检测中，行人尺寸变化大，导致小尺寸行人漏检率高，增加了行人检测的难度。为了降低行人检测漏检率，提高行人检测精度，在级联区域卷积神经网络（cascade regional convolutional neural network，Cascade RCNN）的基础上，将浅层特征与深层特征融合，进行深层特征对浅层特征的特征增强，提高深层信息的利用率，并且增加了一条浅层到深层的通道，将浅层信息直接向上进行传递，提高浅层空间信息的利用率；将行人分类和预测框回归的全连接层改为解耦的回归与分类分支，更加稳健地进行分类和回归整个边界框。在Caltech和ETH行人数据集上进行实验，结果表明，改进的Cascade RCNN与原Cascade RCNN相比，在Caltech行人数据集中大中小尺寸行人漏检率分别降低了7.9个百分点、11.4个百分点和9.1个百分点，平均精度均值提高了3.0个百分点；在ETH行人数据集中漏检率降低了5.6个百分点，平均精度均值提高了2.3个百分点。

关键词: 行人检测, 特征增强, 特征金字塔, 全连接层, 解耦的回归与分类分支

Abstract: In pedestrian detection under complex road conditions, the size of pedestrian varies greatly, and the missing rate of small-size pedestrians is high, which increases the difficulty of pedestrian detection. In order to reduce the missing rate of pedestrian detection, based on Cascade RCNN（cascade regional convolutional neural network）, low-level feature maps are fused with high-level feature maps, which enhances the feature of the low-level feature maps from the high-level feature maps, and improves the utilization of high-level information. It also adds a bottom-up pathway to transmit the low-level feature maps information directly upwards to improve the utilization of low-level feature maps spatial information. The fully connection layer of pedestrian classification and regression of bounding box is changed to the decoupled regression and classification branch, and the more robust classification and regression of the whole bounding box are adopted. This paper conducts experiments on the Caltech and ETH pedestrian datasets. The results show that compared with the original Cascade RCNN, the improved Cascade RCNN reduces the missing rate of large, medium and small pedestrians on the Caltech pedestrian dataset by 7.9 percentage points, 11.4 percentage points and 9.1 percentage points, respectively. The mean average precision is increased by 3.0 percentage points; the missing rate in the ETH pedestrian dataset is reduced by 5.6 percentage points, and the mean average precision is increased by 2.3 percentage points.

Key words: pedestrian detection, feature enhancement, feature pyramid, fully connection layer, decoupled regression and classification branch

刘艳萍, 刘甜. 改进的Cascade RCNN行人检测算法研究[J]. 计算机工程与应用, 2022, 58(4): 229-236.

LIU Yanping, LIU Tian. Improved Cascade RCNN Pedestrian Detection Algorithm Research[J]. Computer Engineering and Applications, 2022, 58(4): 229-236.

参考文献

[1] 蒋家俊.基于卷积神经网络的小目标行人检测研究[D].兰州：兰州理工大学，2018.
JIANG J J.Research on small target pedestrian detection based on convolutional neural network[D].Lanzhou：Lanzhou University of Technology，2018.
[2] 盛智勇，揭真，曲洪权，等.基于改进锚候选框的甚高速区域卷积神经网络的端到端地铁行人检测[J].科学技术与工程，2018，18（22）：95-101.
SHENG Z Y，JIE Z，QU H Q，et al.End-to-end subway pedestrian detection based on very high-speed area convolutional neural network based on improved anchor candidate frame[J].Science Technology and Engineering，2018，18（22）：95-101.
[3] 王飞，王林，张儒良，等.基于融合FPN和Faster R-CNN的行人检测算法[J].数据采集与处理，2019（3）：530-537.
WANG F，WANG L，ZHANG R L，et al.Pedestrian detection algorithm based on fusion of FPN and Faster R-CNN[J].Data Collection and Processing，2019（3）：530-537.
[4] KIM J H，BATCHULUUN G，PARK K R.Pedestrian detection based on Faster R-CNN in nighttime by fusing deep convolutional features of successive images[J].Expert Systems with Applications，2018，114：15-33.
[5] 尚晓航.复杂场景中的行人检测算法研究[D].南京：南京邮电大学，2018.
SAHNG X H.Research on pedestrian detection algorithms in complex scenes[D].Nanjing：Nanjing University of Posts and Telecommunications，2018.
[6] TYCHSENSMITH L，PETERSSON L.Improving object localization with fitness NMS and bounded IoU loss[C]//2018 IEEE Conference on Computer Vision and Pattern Recognition，2018：6877-6885.
[7] HE Y，ZHU C，WANG J，et al.Bounding box regression with uncertainty for accurate object detection[C]//2019 IEEE Conference on Computer Vision and Pattern Recognition，2019：2888-2897.
[8] ZHANG Y，SHEN Y L，ZHANG J.An improved Tiny-YOLOv3 pedestrian detection algorithm[J].Optik-International Journal for Light and Electron Optics，2019，183：17-23.
[9] 张汇，杜煜，宁淑荣，等.基于Faster RCNN的行人检测方法[J].传感器与微系统，2019，38（2）：147-149.
ZHANG H，DU Y，NING S R，et al.Pedestrian detection method based on Faster RCNN[J].Sensors and Microsystems，2019，38（2）：147-149.
[10] GIRSHICK R.Fast R-CNN[C]//Proceedings of the 2015 IEEE International Conference on Computer Vision，2015：1440-1448.
[11] 周滢慜.基于机器视觉的生活垃圾智能分拣系统的设计与实现[D].哈尔滨：哈尔滨工业大学，2018.
ZHOU Y M.Design and implementation of an intelligent sorting system for domestic waste based on machine vision[D].Harbin：Harbin Institute of Technology，2018.
[12] REN S Q，HE K M，GIRSHICK R，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.
[13] LIN T，DOLLAR P，GIRSHICK R，et al.Feature pyramid networks for object detection[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition，2017：936-944.
[14] 宋荣，周大可，杨欣.基于特征融合的尺度感知行人检测[J].电子测量技术，2020，43（5）：116-123.
SONG R，ZHOU D K，YANG X.Scale-aware pedestrian detection based on feature fusion[J].Electronic Measurement Technology，2020，43（5）：116-123.
[15] 李佐龙，王帮海，卢增.多尺度特征融合重建的行人检测方法[J].计算机工程与应用，2021，57（4）：176-182.
LI Z L，WANG B H，LU Z.Pedestrian detection method based on multi-scale feature fusion reconstruction[J].Computer Engineering and Applications，2021，57（4）：176-182.
[16] LIU S，QI L，QIN H F，et al.Path aggregation network for instance segmentation[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition，2018.
[17] GHIASI G，LIN T，LE Q V，et al.NAS-FPN：learning scalable feature pyramid architecture for object detection[C]//2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition，2019：7036-7045.
[18] TAKARLI F，AGHAGOLZADEH A，SEYEDARABI H，et al.Combination of high-level features with low-level features for detection of pedestrian[J].Signal，Image and Video Processing，2016，10（1）：93-101.
[19] SHELHAMER E，LONG J，DARRELL T.Fully convolutional networks for semantic segmentation[J].IEEE Transactions on Pattern Analysis and Machine Intelligence，2017，39（4）：640-651.
[20] WU Y，CHEN Y，YUAN L P，et al.Rethinking classification and localization for object detection[C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition，2020：10183-10192.
[21] CAI Z W，VASCONCELOS N.Cascade R-CNN：high quality object detection and instance segmentation[J].IEEE Transactions on Pattern Analysis and Machine Intelligence，2020，43（5）：1483-1498.
[22] 李伟山，卫晨，王琳.改进的Faster RCNN煤矿井下行人检测算法[J].计算机工程与应用，2019，55（4）：200-207.
LI W S，WEI C，WANG L.Improved Faster RCNN coal mine pedestrian detection algorithm[J].Computer Engineering and Applications，2019，55（4）：200-207.
[23] 胡超超，刘军，张凯，等.基于深度学习的行人和骑行者目标检测及跟踪算法研究[J].汽车技术，2019（7）：19-23.
HU C C，LIU J，ZHANG K，et al.Research on pedestrian and cyclist target detection and tracking algorithm based on deep learning[J].Automotive Technology，2019（7）：19-23.
[24] GUO A X，YIN B Q，ZHANG J，et al.Pedestrian detection via multi-scale feature fusion convolutional neural network[C]//2017 Chinese Automation Congress，2017：1364-1368.
[25] ENZWEILER M，GAVRILA D M.Monocular pedestrian detection：survey and experiments[J].IEEE Transactions on Pattern Analysis and Machine Intelligence，2009，31（12）：2179-2195.
[26] ZHOU P，NI B，GENG C，et al.Scale-transferrable object detection[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition，2018：528-537.
[27] PANG J，CHEN K，SHI J，et al.Libra R-CNN：towards balanced learning for object detection[C]//2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition，2019：821-830.
[28] DOLLAR P，WOJEK C，SCHIELE B，et al.Pedestrian detection：a benchmark[C]//2009 IEEE Conference on Computer Vision and Pattern Recognition，2009：304-311.