DOI: https://doi.org/10.33099/2311-7249/2019-35-2-65-70

CASCADE MODEL OF HYBRID SEGMENTATION FOR AUTOMATIC DECISION OF AEROPHOTO OBJECTS

Serhiy Kovbasiuk, Leonid Kanevskyy, Mykola Romanchuk

Abstract


Current tasks for today are to search and improve the methods of automatic detailed decoding of objects on aerial photographs obtained from unmanned aviation complexes, which would provide sufficient accuracy of detection and recognition of fine-grained objects in the complex topographical conditions of the terrain above which aerial images are obtained. In order to solve this problem in the article an analysis of methods for automatic image processing and models of neural networks built on their basis. From the analysis, a multi-stage conveyor for aerial photographs has been selected, combining detection approaches, elemental segmentation and semantic segmentation for contextualization. Improved models of cascade of segmentation take into account geometric sizes of objects and their correlation, change in scale, conditions of removal. Using the segmentation cascade model for automatic decoding of objects on aerial photos will increase the accuracy of detection and recognition of such objects.


Keywords


neural network, segmentation, aerospace, unmanned aerial systems, cascade of hybrid segmentation

References


H. Zhao, J. Shi, X. Qi, X. Wang, and J. Jia. Pyramid scene parsing network. available at: https://arxiv.org/ abs/1612.01105.pdf.

L.-C. Chen, G. Papandreou, I. Kokkinos, K. Murphy and A. L. Yuille. Semantic image segmentation with deep convolutional nets and fully connected CRFs. available at: https://arxiv.org/abs/ 1412.7062.pdf.

F. Yu, V. Koltun. Multi-scale context aggregation by dilated convolutions. available at: https://arxiv.org/abs/1511.07122.pdf.

T. Xiao, Y. Liu, B. Zhou, Y. Jiang, and J. Sun. Unified perceptual parsing for scene understanding. available at: https://arxiv.org/abs/ 1711.10370.pdf.

T.-Y. Lin, P. Dollár, R. Girshick, K. He, B. Hariharan, S. Belongie. Feature Pyramid Networks for Object Detection. available at: https://arxiv.org/ abs/1612.03144.pdf.

R. Girshick, J. Donahue, T. Darrell and J. Malik. Rich feature hierarchies for accurate object detection and semantic segmentation. available at: https://arxiv.org/abs/1311.2524.pdf.

J. R. Uijlings, K. E. V. D. Sande, T. Gevers, and A. W. Smeulders. Selective search for object recognition. available at: https://arxiv.org/abs/1807.05511.pdf.

J. Dai, K. He, Y. Li, S. Ren, J. Sun. Instance-sensitive fully convolutional networks. available at: https://arxiv.org/abs/1603.08678.pdf.

Q. Haozhi, D. Jifeng, J. Xiangyang , W. Yichen. Fully Convolutional Instance-aware Semantic Segmentation. available at: https://arxiv.org/abs/1611.07709.pdf.

J. Hosang, R. Benenson, P. Doll´ar, B. Schiele. What makes for effective detection proposals? available at: https://arxiv.org/abs/1502.05082.pdf.

J. Long, E. Shelhamer, T. Darrell. Fully convolutional networks for semantic segmentation. available at: https://arxiv.org/abs/ 1411.4038.pdf.

K. He, G. Gkioxari, P. Doll´ar, R. Girshick. Mask R-CNN. available at: https://arxiv.org/abs/ 1703.06870.pdf.

L.-C. Chen, G. Papandreou, I. Kokkinos, K. Murphy, A. L. Yuille. DeepLab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected CRFs. available at: https://arxiv.org/abs/1606.00915.pdf.

S. Ren, K. He, R. Girshick, J. Sun. Faster R-CNN: Towards realtime object detection with region proposal networks. available at: https://arxiv.org/abs/1506.01497.pdf.

J. Tighe, M. Niethammer, S. Lazebnik. Scene parsing with object instances and occlusion ordering. available at: https://arxiv.org/abs/1806.03772.pdf.

Z. Tu, X. Chen, A. L. Yuille, S.-C. Zhu. Image parsing: Unifying segmentation, detection, and recognition. available at: https://arxiv.org/ abs/0502172.pdf.

J. Yao, S. Fidler, R. Urtasun. Describing the scene as a whole: Joint object detection, scene classification and semantic segmentation. available at: https://arxiv.org/abs/1207.0372.pdf.

M. Sun, B.-S. Kim, P. Kohli, S. Savarese. Relating things and stuff via object property interactions. available at: http://svl.stanford.edu /assets/papers/pami14_acrf.pdf.

J. Yao, S. Fidler, and R. Urtasun. Describing the scene as a whole: Joint object detection, scene classification and semantic segmentation. available at: https://ttic.uchicago.edu/~rurtasun/publications/ yaoetal_cvpr12.pdf.

S. Zhang, L. Wen, X. Bian, Z. Lei, S. Z. Li. Single-shot refinement neural network for object detection. available at: https://arxiv.org/abs/ 1711.06897.pdf.

J. Redmon, S. Divvala, R. Girshick, A. Farhadi. You Only Look Once: Unified, Real-Time Object Detection. available at: https://arxiv.org/abs/ 1506.02640.pdf.

S. Gidaris, N. Komodakis. Object detection via a multiregion and semantic segmentation-aware CNN model. available at: https://arxiv.org/abs/ 1711.06897.pdf.

Z. Cai, N. Vasconcelos. Cascade R-CNN: Delving into high quality object detection. available at: https://arxiv.org/ abs/1712.00726.pdf.

Z. Cai, N. Vasconcelos. Aggregated residual transformations for deep neural networks. available at: https://arxiv.org/abs/ 1611.05431.pdf.

J. Dai, H. Qi, Y. Xiong, Y. Li, G. Zhang, H. Hu, Y. Wei. Deformable convolutional networks. available at: https://arxiv.org/abs/1703.06211.pdf.

K. Chen, J. Pang, J. Wang and other. Hybrid Task Cascade for Instance Segmentation. available at: https://arxiv.org/abs/ 1901.07518.pdf.

T.-Y. Lin, P. Goyal, R. Girshick, K. He, P. Dollár. Focal Loss for Dense Object Detection. available at: https://arxiv.org/abs/1708.02002.pdf


GOST Style Citations


1. Zhao H. Pyramid scene parsing network [Електронний ресурс] / H. Zhao, J. Shi, X. Qi, X. Wang, and J. Jia // Режим доступу : https://arxiv.org/abs/1612.01105.pdf.

2. Chen L.-C. Semantic image segmentation with deep convolutional nets and fully connected CRFs [Електронний ресурс] / L.-C. Chen, G. Papandreou, I. Kokkinos, K. Murphy and A. L. Yuille // Режим доступу : https://arxiv.org/abs/1412.7062.pdf.

3. Yu F. Multi-scale context aggregation by dilated convolutions [Електронний ресурс] / F. Yu, V. Koltun // Режим доступу : https://arxiv.org/abs/1511.07122.pdf.

4. Xiao T. Unified perceptual parsing for scene understanding [Електронний ресурс] / T. Xiao, Y. Liu, B. Zhou, Y. Jiang, and J. Sun // Режим доступу : https://arxiv.org/abs/1711.10370.pdf.

5. Lin T.-Y. Feature Pyramid Networks for Object Detection [Електронний ресурс] / T.-Y. Lin, P. Dollár, R. Girshick, K. He, B. Hariharan, S. Belongie // Режим доступу : https://arxiv.org/abs/1612.03144.pdf.

6. Girshick R. Rich feature hierarchies for accurate object detection and semantic segmentation [Електронний ресурс] / R. Girshick, J. Donahue, T. Darrell and J. Malik // Режим доступу : https://arxiv.org/abs/1311.2524.pdf.

7. Uijlings J. R. Selective search for object recognition [Електронний ресурс] / J. R. Uijlings, K. E. V. D. Sande, T. Gevers, and A. W. Smeulders // Режим доступу : https://arxiv.org/ abs/1807.05511.pdf.

8. Dai J. Instance-sensitive fully convolutional networks [Електронний ресурс] / J. Dai, K. He, Y. Li, S. Ren, J. Sun // Режим доступу : https://arxiv.org/abs/1603.08678.pdf.

9. Li Y. Fully Convolutional Instance-aware Semantic Segmentation [Електронний ресурс] / Q. Haozhi , D. Jifeng , J. Xiangyang , W. Yichen // Режим доступу : https://arxiv.org/abs/1611.07709.pdf.

10. Hosang J. What makes for effective detection proposals? [Електронний ресурс] / J. Hosang, R. Benenson, P. Doll´ar, B. Schiele // Режим доступу : https://arxiv.org/abs/1502.05082.pdf.

11. Long J. Fully convolutional networks for semantic segmentation [Електронний ресурс] / J. Long, E. Shelhamer, T. Darrell // Режим доступу : https://arxiv.org/abs/1411.4038.pdf.

12. He K. Mask R-CNN [Електронний ресурс] / K. He, G. Gkioxari, P. Doll´ar, R. Girshick // Режим доступу : https://arxiv.org/abs/1703.06870.pdf.

13. Chen L.-C. DeepLab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected CRFs [Електронний ресурс] / L.-C. Chen, G. Papandreou, I. Kokkinos, K. Murphy, A. L. Yuille // Режим доступу : https://arxiv.org/abs/1606.00915.pdf.

14. Ren S. Faster R-CNN: Towards realtime object detection with region proposal networks [Електронний ресурс] / S. Ren, K. He, R. Girshick, J. Sun // Режим доступу : https://arxiv.org/abs/1506.01497.pdf.

15. Tighe J. Scene parsing with object instances and occlusion ordering [Електронний ресурс] / J. Tighe, M. Niethammer, S. Lazebnik // Режим доступу : https://arxiv.org/abs/ 1806.03772.pdf.

16. Tu Z. Image parsing: Unifying segmentation, detection, and recognition [Електронний ресурс] / Z. Tu, X. Chen, A. L. Yuille, S.-C. Zhu // Режим доступу : https://arxiv.org/abs/0502172.pdf.

17. Yao J. Describing the scene as a whole: Joint object detection, scene classification and semantic segmentation [Електронний ресурс] / J. Yao, S. Fidler, R. Urtasun // Режим доступу : https://arxiv.org/abs/1207.0372.pdf.

18. Sun M. Relating things and stuff via object property interactions [Електронний ресурс] / M. Sun, B.-S. Kim, P. Kohli, S. Savarese // Режим доступу : http://svl.stanford.edu/assets/papers/pami14_acrf.pdf.

19. Yao J. Describing the scene as a whole: Joint object detection, scene classification and semantic segmentation [Електронний ресурс] / J. Yao, S. Fidler, and R. Urtasun // Режим доступу : https://ttic.uchicago.edu/~rurtasun/ publications/ yao_et_al_cvpr12.pdf.

20. Zhang S. Single-shot refinement neural network for object detection [Електронний ресурс] / S. Zhang, L. Wen, X. Bian, Z. Lei, S. Z. Li // Режим доступу : https://arxiv.org/abs/ 1711.06897.pdf.

21. Redmon J. You Only Look Once: Unified, Real-Time Object Detection [Електронний ресурс] / J. Redmon, S. Divvala, R. Girshick, A. Farhadi // Режим доступу : https://arxiv.org/abs/1506.02640.pdf.

22. Gidaris S. Object detection via a multiregion and semantic segmentation-aware CNN model [Електронний ресурс] / S. Gidaris, N. Komodakis // Режим доступу : https://arxiv.org/abs/1711.06897.pdf.

23. Cai Z. Cascade R-CNN: Delving into high quality object detection [Електронний ресурс] / Z. Cai, N. Vasconcelos // Режим доступу : https://arxiv.org/abs/1712.00726.pdf.

24. Xie S. Aggregated residual transformations for deep neural networks [Електронний ресурс] / Z. Cai, N. Vasconcelos // Режим доступу : https://arxiv.org/abs/1611.05431.pdf.

25. Dai J. Deformable convolutional networks [Електронний ресурс] / J. Dai, H. Qi, Y. Xiong, Y. Li, G. Zhang, H. Hu, Y. Wei // Режим доступу : https://arxiv.org/abs/ 1703.06211.pdf.

26. Chen K. Hybrid Task Cascade for Instance Segmentation [Електронний ресурс] / K. Chen, J. Pang, J. Wang and other // Режим доступу : https://arxiv.org/abs/ 1901.07518.pdf.

27. Lin T.-Y. Focal Loss for Dense Object Detection [Електронний ресурс] / T.-Y. Lin, P. Goyal, R. Girshick, K. He, P. Dollár // Режим доступу : https://arxiv.org/abs/1708.02002.pdf. 





ISSN 2410-7336 (Online)

ISSN 2311-7249 (Print)