基于MaskR-CNN的各类主干网络应用差异分析

朱 圣果, 李 丹
四川大学锦城学院

摘要

随着基于卷积神经网络的R-CNN在VOC2012数据集上较此前最佳结果在平均精度（mAP）上获得了30%以上的提升，达到53.3%mAP[1]后，深度学习算法自此正式步入目标检测领域，凭借较强的通用性和特征迁
移能力、高精确度、低优化及维护成本等优势迅速取代传统目标检测算法，逐步发展出FastR-CNN[2]、FasterR-CNN[3]、
MaskR-CNN[4]等一系列衍生，此间SPP（SpatialPyramidPooling）、RoIPooling（RegionofInterestPooling）、
multi-taskloss、RPN（RegionProposalNetwork）、FPN（FeaturePyramidNetwork）、RoiAlign[4]等概念的提出为根
据实际应用场景调整主体算法跟主干网络结构提供了基础。

关键词

MaskR-CNN；实例分割；backbone；ResNet；FBNet

参考

[1]R. Girshick, J. Donahue, T. Darrell, and J. Malik. Rich feature hierarchies for accurate object detection and semantic segmentation. arXiv preprint arXiv:1311.2524, 2013.

[2]Girshick R . Fast R-CNN[J]. Computer Science, 2015.

[3]Ren S , He K , 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, 2015, 39(6).

[4]Kaiming H , Georgia G , Piotr D , et al. Mask R-CNN[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2018:1-1.

[5]He K , Zhang X , Ren S , et al. Spatial Pyramid Pooling in Deep Convolutional Networks for Visual Recognition[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2014, 37(9):1904-16.

[6]Lin T Y , Dollár, Piotr, Girshick R , et al. Feature Pyramid Networks for Object Detection[J]. 2016.

[7] 朱福喜 , 汤怡群 , 傅建明 . 人工智能原理 [M]. 武汉 : 武汉大 学出版社 , 2002.

[8]Campblls M , Jr A J H , Hsu F H . Deep Blue[J]. Artificial intelligence, 2002, 134(1/2).

[9]Hinton G , Salakhutdinov R . Reducing the dimensionality of data with neural networks[J]. Science, 2006, 313(5786):p. 504-507.

[10]Lowe D G . Distinctive Image Features from Scale-Invariant Keypoints[J]. International Journal of Computer Vision, 2004, 60(2):91-110.

[11]DALAL,N. Histgrams of oriented gradients for human detection[J]. proc of cvpr, 2005.

[12]Felzenszwalb P F , Mcallester D A , Ramanan D . A Discriminatively Trained, Multiscale, Deformable Part Model[C]// 2008 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR 2008), 24-26 June 2008, Anchorage, Alaska, USA. IEEE, 2008.

[13] 周晓彦 , 王珂 , 李凌燕 . 基于深度学习的目标检测算法综述 [J]. 电子测量技术 , 2017(11):89-93.

[14]He K , Zhang X , Ren S , et al. Deep Residual Learning for Image Recognition[J]. 2015.

[15]Wu B , Dai X , Zhang P , et al. FBNet: Hardware-Aware Efficient ConvNet Design via Differentiable Neural Architecture Search[J]. 2018.

[16]Zoph B , Le Q V . Neural Architecture Search with Reinforcement Learning[J]. 2016.

[17] 刘丽 , 匡纲要 . 图像纹理特征提取方法综述 [J]. 中国图象图 形学报 , 2009(04):63-76.