Crack Detection with Multi-task Enhanced Faster R-CNN Model

To solve the low accuracy problem when dealing with multiple or small objects detection, a Multi-task Enhanced crack detection based on Faster R-CNN (ME-Faster R-CNN) was proposed in the dam safety monitoring. MEFaster R-CNN adopts the Multi-source Adaptive Balance TrAdaBoost based on K-means (K-MABtrA) to enhance the small-scale samples. ME-Faster R-CNN uses RestNet-50 network to extract feature and apply multi-task enhanced RPN module to generate the candidate regions through adopting the appropriate size and dimension of anchor box. The features map and candidate regions are processed to detect the dam cracks. Experimental results demonstrate that the proposed ME-Faster R-CNN with K-MABtrA method can obtain 82.52% average IoU and 80.02% average precision mAP, respectively. Compared with Faster R-CNN with the same parameters, the average IoU and mAP can increase 1.06% and 1.56%, respectively.

[1]  Jian Sun,et al.  Deep Residual Learning for Image Recognition , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[2]  Trevor Darrell,et al.  Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation , 2013, 2014 IEEE Conference on Computer Vision and Pattern Recognition.

[3]  Ross B. Girshick,et al.  Fast R-CNN , 2015, 1504.08083.

[4]  Andrew Zisserman,et al.  Very Deep Convolutional Networks for Large-Scale Image Recognition , 2014, ICLR.

[5]  Rob Fergus,et al.  Visualizing and Understanding Convolutional Networks , 2013, ECCV.

[6]  Jordan M. Malof,et al.  Improving convolutional neural networks for buried target detection in ground penetrating radar using transfer learning via pretraining , 2017, Defense + Security.

[7]  Wei Liu,et al.  SSD: Single Shot MultiBox Detector , 2015, ECCV.

[8]  Ali Farhadi,et al.  You Only Look Once: Unified, Real-Time Object Detection , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[9]  Ali Farhadi,et al.  YOLOv3: An Incremental Improvement , 2018, ArXiv.

[10]  Ali Farhadi,et al.  YOLO9000: Better, Faster, Stronger , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[11]  Luc Van Gool,et al.  Efficient Non-Maximum Suppression , 2006, 18th International Conference on Pattern Recognition (ICPR'06).

[12]  Jing Yu,et al.  Classification of thyroid nodules in ultrasound images using deep model based transfer learning and hybrid features , 2017, 2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

[13]  Walter Kellermann,et al.  Efficient target activity detection based on recurrent neural networks , 2017, 2017 Hands-free Speech Communications and Microphone Arrays (HSCMA).

[14]  Kaiming He,et al.  Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks , 2015, IEEE Transactions on Pattern Analysis and Machine Intelligence.