Fast Ship Detection in Optical Remote Sensing Images Based on Sparse MobileNetV2 Network
- 24 Downloads
Ship detection in optical remote sensing images (ORSIs) has drawn lots of attention because of its extensive potential in maritime applications. Although many methods have been proposed in recent years, there are still great challenges for improving the detection accuracy and detection speed. In this paper, a fast ship detection method in optical remote sensing images based on sparse MobileNetV2 network is proposed, which has high accuracy and fast detection speed. Ship detection problem is turned into a sub-image classification one, which successfully avoids the massive computation caused by the region proposal stage in previous methods. The sparse MobileNetV2 network has high detection accuracy and less computation benefited from the convolutional neural networks and the depth separable convolution. Furthermore, the pruning method is used to compress the network to decrease model complexity and prevent overfitting. Several experiments are conducted based on some optical remote sensing images from Google Earth. The results demonstrate that the proposed method achieves over 5x speed enhancement compared with several mainstream ship detection methods, while the accuracy is competitive.
KeywordsOptical remote sensing image Ship detection Sparse MobileNetV2 network Model compression
- 3.Ji-yang, Y., Dan, H., Lu-yuan, W., et al.: A real-time on-board ship targets detection method for optical remote sensing satellite. In: 2016 IEEE 13th International Conference on Signal Processing (ICSP), pp. 204–208. IEEE (2016)Google Scholar
- 7.Redmon, J., Divvala, S., Girshick, R., et al.: You only look once: unified, real-time object detection. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 779–788 (2016)Google Scholar
- 8.Liu, W., Anguelov, D., Erhan, D., et al.: SSD: single shot multibox detector. In: European Conference on Computer Vision, pp. 21–37 (2016)Google Scholar
- 10.Sandler, M., Howard, A., Zhu, M., et al.: MobileNetV2: inverted residuals and linear bottlenecks. In: 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 4510–4520 (2018)Google Scholar