Ship detection in SAR Image using YOLOv2

Ship detection in SAR images is a challenge and has traditionally been carried out using pixel based algorithms such as CFAR, in this paper we use a deep learning based algorithm called YOLOv2 for the aforementioned task and test its performance on three datasets, at different resolution and quality, with two datasets called DS1 and DS2 consisting of over 400 high resolution SAR images with a high ship to image ratio, in which DS1 consists of raw, unfiltered images and the third dataset called DS3 which consists of full scale 5500 single ship images of lower resolution. High detection results are observed which further improve once we train the network by applying transfer learning using weights from DS3 by truncating the last layer of the network and continue training on the combined first two datasets.

[1]  Tianqi Chen,et al.  Empirical Evaluation of Rectified Activations in Convolutional Network , 2015, ArXiv.

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

[3]  Björn Tings,et al.  Ship-Iceberg Discrimination with Convolutional Neural Networks in High Resolution SAR Images , 2016 .

[4]  H. Zhang,et al.  An improved nonparametric CFAR method for ship detection in single polarization synthetic aperetuer radar imagery , 2016, 2016 IEEE International Geoscience and Remote Sensing Symposium (IGARSS).

[5]  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.

[6]  Mingsheng Liao,et al.  Ship Detection in SAR Image Based on the Alpha-stable Distribution , 2008, Sensors.

[7]  Yoshua Bengio,et al.  How transferable are features in deep neural networks? , 2014, NIPS.

[8]  Jianwei Li,et al.  Ship detection in SAR images based on an improved faster R-CNN , 2017, 2017 SAR in Big Data Era: Models, Methods and Applications (BIGSARDATA).

[9]  Liang Chen,et al.  An Intensity-Space Domain CFAR Method for Ship Detection in HR SAR Images , 2017, IEEE Geoscience and Remote Sensing Letters.

[10]  H. Enayati,et al.  Automatic Ship Detection in Single-Pol SAR Images Using Texture Features in Artificial Neural Networks , 2015 .

[11]  Jitendra Malik,et al.  Region-Based Convolutional Networks for Accurate Object Detection and Segmentation , 2016, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[12]  Vishal M. Patel,et al.  SAR Image Despeckling Using a Convolutional Neural Network , 2017, IEEE Signal Processing Letters.

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

[14]  Zhao Lin,et al.  A modified faster R-CNN based on CFAR algorithm for SAR ship detection , 2017, 2017 International Workshop on Remote Sensing with Intelligent Processing (RSIP).