臺灣博碩士論文加值系統

本論文提出一個深度學習單階段實例分割演算法，藉由本演算法預測影像中物件的遮罩形狀，並根據獲取的資料來進行魚體資訊分析。相較於傳統物件辨識預測物件的邊界框方式，實例分割演算法是預測畫面中物件的遮罩，本演算法以單階段的實例分割演算法SOLOV2為基礎，藉由將演算法Backbone的部分替換為深度匯集層，藉由深度匯集層本身擷取特徵的強項來加強所提出的單階段實例分割演算法結果，相對於SOLOV2特徵擷取部分的ResNet，深度匯集層藉由使用複雜的匯集架構，來取得更詳細的特徵，所建立的演算法可以因此分割出更詳細的結果。
演算法使用COCO資料集進行訓練與測試，將測試結果與SOLOV2進行比較、分析，結果顯示本論文提出的實例分割演算法AP略低於SOLOV2，然而因為深度匯集層可以取得詳細的特徵，在分割結果圖中可以看到本論文提出的實例分割演算法分割的結果更好，比SOLOV2的結果更加詳細。在應用上本論文提出的實例分割演算法可以應用於養殖漁業中，藉由演算法分割出魚隻，並使用主成分分析的方法取得魚隻資訊，協助養殖業者進行養殖規劃。

The purpose of this thesis is develop a deep learning single-stage instance segmentation algorithm, which predicts the mask shape of objects in the image by this algorithm, and analyzes the fish information according to the obtained data. Compared with object detection method, instance segmentation algorithm is to obtain the object mask. Our proposed algorithm is based on the single-stage instance segmentation algorithm SOLOV2, by replacing Backbone. For the deep layer aggregation, our proposed single-stage instance segmentation algorithm results are enhanced by the advantage of feature extraction.
Our algorithm is trained and tested on COCO dataset, compared and analyzed with the original SOLOV2, and the results shows the AP of our method a little less than SOLOV2 Because the ability of deep layer aggregation extracting better feature, the result image shows better rather than SOLOV2. In applications, our method can be used to get fish information by principal component analysis, to assist aquaculture operators in breeding planning.

摘要 I
Abstract II
目次 III
圖目次 V
表目次 VI
第一章 緒論 1
1.1 研究動機 1
1.2 研究目的 2
1.3 論文架構 3
第二章 相關研究探討 4
2.1 特徵擷取網路 4
2.2 實例分割演算法 6
第三章 研究方法 8
3.1 演算法架構 8
3.2 深度匯集層結構 8
3.2.1 迭代式深度匯集 9
3.2.2 階層式深度匯集 9
3.2.3 深度匯集層 11
3.3 SOLOV2影像分割演算法 11
3.3.1 Backbone 12
3.3.2 Neck 12
3.3.3 Head 13
3.3.4 損失函數 14
第四章 實驗結果與比較 16
4.1 實驗平台 16
4.2 資料集 16
4.3 實驗評估指標 16
4.4 實例分割演算法實驗結果與比較 18
4.5實例分割應用於魚體資訊分析 21
4.6 分魚機系統 26
4.6.1 非侵入式魚隻資訊測量系統 27
4.6.2 魚隻大小分類模組 28
4.6.3 分魚機系統其他效益 29
第二章 結論與未來展望 31
5.1 結論 31
5.2 未來展望 32
參考文獻 33

[1] X. Wang, R. Zhang, T. Kong, L. Li, and C. Shen, “SOLOv2: Dynamic and Fast Instance Segmentation,” in Neural Information Processing Systems, no. 1487, pp. 17721-17732, Dec. 2020.
[2] H. Abdi, and L. J. Williams, “Principal Component Analysis.” in Wiley Interdisciplinary Reviews: Computational Statistics 2, no. 4, pp. 433-459, July 2010.
[3] F. Yu, D. Wang, E. helhamer, and T. Darrell, “Deep Layer Aggregation,” in IEEE Conference on Computer Vision and Pattern Recognition, pp. 2403-2412, Dec. 2018.
[4] K. He, X. Zhang, S. Ren, and J. Sun, “Identity Mappings in Deep Residual Networks,” in European Conference on Computer Vision, pp. 630-645, Sep. 2016.
[5] G. Huang, Z. Liu, K. Q. Weinberger, and L. V. D. Maaten, “Densely Connected Convolutional Networks,” in IEEE Conference on Computer Vision and Pattern Recognition, pp. 2261-2269, Nov. 2017.
[6] K. Simonyan, and A. Zisserman, “Very Deep Convolutional Networks for Large-Scale Image Recognition,” in International Conference on Learning Representations, May 2015.
[7] O. Ronneberger, P. Fischer, and T. Brox, “U-Net: Convolutional Networks for Biomedical Image Segmentation,” in International Conference on Medical Image Computing and Computer-Assisted Intervention, pp. 234-241, Nov. 2015.
[8] E. Shelhamer, J. Long, and T. Darrell, “Fully Convolutional Networks for Semantic Segmentation,” in IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 39, no. 4, pp. 640-651, Apr. 2017.
[9] K. He, G. Gkioxari, P. Dollár, and R. Girshick, “Mask R-CNN,” in IEEE International Conference on Computer Vision, pp. 2980-2988, Oct. 2017.
[10] S. Ren, K. He, R. Girshick, and J. Sun, “Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks,” in IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 39, no. 6, pp. 1137-1149, June 2017.
[11] S. Liu, L. Qi, H. Qin, J. Shi, and J. Jia, “Path Aggregation Network for Instance Segmentation,” in IEEE Conference on Computer Vision and Pattern Recognition, pp. 8759-8768, June 2018.
[12] X. Wang, T. Kong, C. Shen, Y. Jiang, and L. Li, “SOLO: Segmenting Objects by Locations,” in European Conference on Computer Vision, pp. 649-665, Aug. 2020.
[13] R. Liu, J. Lehman, P. Molino, F. P. Such, E. Frank, A. Sergeev, and J. Yosinski, “An Intriguing Failing of Convolutional Neural Networks and the CoordConv Solution.” in Neural Information Processing Systems, vol. 18, pp. 9628–9639, July 2018.
[14] T.-Y. Lin, P. Goyal, R. Girshick, K. He, and P. Dollár, “Focal Loss for Dense Object Detection,” in IEEE Trans. Pattern Anal. Mach. Intell., vol. 42, pp. 318-327, Feb. 2020.
[15] F. Milletari, N. Navab and S. A. Ahmadi, “V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentation,” in Fourth International Conference on 3D Vision, pp. 565-571, Oct. 2016.
[16] K. Chen, K. Chen, J. Wang, J. Pang, Y. Cao, Y. Xiong, X. Li, S. Sun, W. Feng, Z. Liu, J. Xu, Z. Zhang, D. Cheng, C. Zhu, T. Cheng, Q. Zhao, B. Li, X. Lu, R. Zhu, Y. Wu, J. Dai, J. Wang, J. Shi, W. Ouyang, C. C. Loy, and D. Lin, “MMDetection: Open MMLab Detection Toolbox and Benchmark, ” in arXiv preprint arXiv:1906.07155, June 2019.
[17] A. Paszke, S. Gross, F. Massa, A. Lerer, J. Bradbury, G. Chanan, T. Killeen, Z. Lin, N. Gimelshein, L. Antiga, A. Desmaison, A. Kopf, E. Yang, Z. DeVito, M. Raison, A. Tejani, S. Chilamkurthy, B. Steiner, L. Fang, J. Bai, and S. Chintala, “Pytorch: An Imperative Style High-Performance Deep Learning Library,” in Neural Information Processing Systems, vol. 33, no. 721, pp. 8026-8037, Dec. 2019.
[18] COCO:: T.-Y. Lin, M. Maire, S. Belongie, J. Hays, P. Perona, D. Ramanan, P. Dollár, and C. L. Zitnick, “Microsoft coco: Common objects in context,” in European Conference on Computer Vision, pp. 740-755, 2014.
[19] B. C. Russell, A. Torralba, K. P. Murphy, and W. T. Freeman, “LabelMe:A Database and Web-Based Tool for Image Annotation,” in International Journal of Computer Vision, vol. 77, no. 1, pp. 157-173, May 2008.