(3.235.108.188) 您好!臺灣時間:2021/02/28 00:05
字體大小: 字級放大   字級縮小   預設字形  
回查詢結果

詳目顯示:::

我願授權國圖
: 
twitterline
研究生:黃羿豪
研究生(外文):HUANG, YI-HAO
論文名稱:深度學習光影強固之嵌入式物件精密實時定位操控技術
論文名稱(外文):Deep-trained illumination-robust precision positioning for real-time manipulation of embedded objects
指導教授:李志鴻李志鴻引用關係
指導教授(外文):LI, CHIH-HUNG
口試委員:李仕宇連震杰李志鴻
口試委員(外文):LI, SHIH-YULIEN, JENN-JIERLI, CHIH-HUNG
口試日期:2020-07-14
學位類別:碩士
校院名稱:國立臺北科技大學
系所名稱:製造科技研究所
學門:工程學門
學類:機械工程學類
論文種類:學術論文
論文出版年:2020
畢業學年度:108
語文別:中文
論文頁數:65
中文關鍵詞:深度學習影像處理卷積神經網路生成對抗網路手臂定位One-Shot
外文關鍵詞:Deep LearningComputer visionConvolutional neural networkGenerative adversarial networkPix2PixLocalizationData augmentationOne-Shot
相關次數:
  • 被引用被引用:0
  • 點閱點閱:28
  • 評分評分:系統版面圖檔系統版面圖檔系統版面圖檔系統版面圖檔系統版面圖檔
  • 下載下載:0
  • 收藏至我的研究室書目清單書目收藏:0
光影效果通常會導致視覺辨識或是定位的錯誤或失敗。儘管基於ConvNet的視覺辨識定位系統對光影汙染效果顯示出了強大的適應性,但是在某些強烈的光影汙染效果下,辨識定位系統仍可能會失敗。在本研究中,我們提出了一種利用 Pix2Pix GAN的數據擴增方法,用於在各種光影汙染效果下自動生成對像圖像。在對視覺辨識定位ConvNet進行訓練時,將生成的光影效果圖像包括在內,用以豐 富訓練集,從而在強烈光影效果下具有更好的辨識及定位的性能。實驗 數 據證明,物件 座 標辨識的準確性大大提高。所提出的框架亦保持了我們所開發的「 One-Shot」 概 念,意義是用戶只需要對目標物體拍照 一次 即可完成一套對光影有強大適應性的物件座標辨識器 。
Illumination effects often result in errors or failure in visual object localization. Whereas ConvNet-based object localization frameworks have shown tremendous robustness to the illumination effect, under some strong illumination effects, the system may still fail. In this paper, the author proposes a data augmentation method utilizing Pix2Pix GAN for automatic generation of object images under various illumination effects. Upon training for the object localization ConvNet, the generated images are included to enrich the training set for a better performance under strong illumination effects. Experimental evidence shows that the accuracy of object coordinate detection can be improved significantly. The proposed framework maintains our concept of “One-Shot” where the user only needs to take a basis photo of the target object in order to create an illumination-robust object coordinate detector.
目錄
摘要 i
ABSTRACT ii
誌謝 iv
目錄 v
表目錄 vii
圖目錄 viii
第一章 緒論 1
1.1 前言 1
1.2 研究動機 2
1.3 研究目標 3
1.4 論文架構 4
第二章 文獻探討 6
2.1 深度卷積神經網路位置辨識 6
2.2 One-Shot架構 6
2.3 生成對抗網路 8
2.3.1 傳統GAN 8
2.3.2 Pix2Pix GAN簡介 11
第三章 方法驗證實驗 12
3.1 實驗目的與規劃 12
3.1.1 拍攝光影影像 12
3.1.2 Pix2Pix GAN訓練 14
3.1.3 One-Shot訓練 16
3.2 光影污染位置辨識實驗 17
3.3 驗證標準 19
3.4 實驗結果總結 26
第四章 強烈光影生成實驗 29
4.1 實驗目的與規劃 29
4.1.1 拍攝光影影像 29
4.1.2 Pix2Pix GAN訓練 32
4.1.3 One-Shot II訓練 38
4.2 光影污染位置辨識實驗 39
4.3 定位驗證標準 43
4.4 靜態定位實驗結果 49
4.5 真實vs生成影像準確度比較 51
4.6 不同光影強度下準確度比較 52
第五章 手臂控制實驗 53
5.1 實驗設置 53
5.2 影像定位手臂控制實驗 54
5.3 驗證標準 57
5.4 實驗結果 58
第六章 結論與未來展望 63
參考文獻 64


參考文獻
1. R. Jain, R. Kasturi, B. G. Schunck, “Machine Vision”, McGraw-Hill,1995
2. L. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio, “Generative adversarial nets”, Advances in Neural Infor-mation Processing Systems 27 (NIPS), 2014
3. A. Krizhevsky, I. Sutskever, and G. E. Hinton, “ImageNet Classification with Deep Con-volutional Neural Networks”, Neural Information Processing Systems, 2012
4. C.H.G. Li, Y.M. Chang, “Automated visual positioning and precision placement of a workpiece using deep learning”. Int J Adv Manuf Technol 104, 2019, pp. 4527-4538
5. Y.M. Chang, C.H.G. Li, and Y.F. Hong, “Real-Time Object Coordinate Detection and Manipulator Control Using Rigidly Trained Convolutional Neural Networks”, 2019 IEEE 15th International Conference on Automation Science and Engineering (CASE 2019), Van-couver, BC, Canada, 2019
6. Y.H. Huang, Y.M. Chang, C.H.G. Li, “Illumination-robust object coordinate detection by adopting Pix2Pix GAN for training image generation”, International conference on technologies and applications of artificial intelligence (TAAI), 2019
7. P. Isola, J.Y. Zhu, T. Zhou, A. A. Efros, “Image-To-Image Translation with Conditional Adversarial Networks”, The IEEE Conference on Computer Vision and Pattern Recogni-tion (CVPR), 2017, pp. 1125-1134.
8. N. Sunderhauf, S. Shirazi, F. Dayoub, B. Upcroft, and M. Milford, “On the performance of ConvNet features for place recognition”, IEEE International conference on intelligent robots and systems (IROS), 2015
9. R. Girshick, “Fast R-CNN”. The IEEE International conference on computer vision (ICCV), 2015
10. M. Oquab, L. Bottou, I. Laptev, J. Sivic, “Is object localization for free? - weakly-super-vised learning with convolutional neural networks”. IEEE Conference on computer vision and pattern recognition (CVPR), 2015
11. S. Ren, K. He, R. Girshick, J. Sun, “Faster R-CNN: towards real time object detection with region proposal networks”. Advances inneural information processing systems (NIPS), 2015
12. J. Redmon, S. Divvala, R. Girshick, A. Farhadi, “You Only Look Once: Unified, Real-Time Object Detection”, The IEEE Conference on Computer Vision and Pattern Recogni-tion (CVPR), 2016, pp. 779-788
13. A. D'Innocente, F. M. Carlucci, M. Colosi, B. Caputo, “Bridging between Computer and Robot Vision through Data Augmentation: a Case Study on Object Recognition”, Inter-national Conference on Computer Vision Systems(ICVS), 2017, pp. 384-393
14. R. Girshick, J. Donahue, T. Darrell, J. Malik, “Rich feature hierarchies for accurate object detection and semantic segmentation”, The IEEE Conference on Computer Vision and Pat-tern Recognition (CVPR), 2014, pp. 580-587
15. P. Isola, J. Y. Zhu, T. Zhou, A. A. Efros, “Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks”, The IEEE International Conference on Com-puter Vision (ICCV), 2017, pp. 2223-2232
電子全文 電子全文(網際網路公開日期:20250810)
QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top
系統版面圖檔 系統版面圖檔