臺灣博碩士論文加值系統

隨著科技的不斷進步，自動駕駛車成為了眾所矚目的研究領域。而在目前自動駕駛
的領域中，行走的決策很大一部份是使用電腦視覺來進行，如車道維持輔助系統(Lane
Keeping Aid, LKA)、車道變換輔助系統（Audi Side Assist）、主動式車距調節巡航系統
（Adaptive Cruise Control, ACC）等功能，都須建立在一個關鍵的技術上，此技術名為駕
駛視覺感知(Driving Visual Perception)，其中包含靜態的車道線辨識與可行駛區域的判定
和動態的交通場景(汽車、機車、行人等)，因為它們能夠幫助車輛了解路面狀況，進而
決定安全行駛路線。
在本文中，參考了 YOLOP 這一駕駛視覺感知模型，並對其進行了改良。前後調整
了兩個模塊和激勵函數的選擇，使模型的精準度和運算速度得到了提升。本論文提出的
方法在 BDD100K 這個大型且具有挑戰性的數據集上也獲得了出色的結果，相較於原始
的 YOLOP 無論是精準度還是速度在各指標上都獲得優異的結果，在可行駛區域檢測的
實驗中準確度達到 97.6%；在車道線檢測的實驗中準確度達到 73.6%；在交通物件檢測
的實驗中召回率達到 88.3%平均準確度達到 76.9%；速度方面達到 87.7FPS 相較於原始
YOLOP 提升了 51.7%的 FPS。

With the continuous advancement of technology, autonomous driving has become a
highly prominent research field. In the current domain of autonomous driving, a significant part
of the decision-making process relies on computer vision, including functionalities such as Lane
Keeping Aid (LKA), Audi Side Assist, and Adaptive Cruise Control (ACC). These systems
heavily rely on a crucial technology known as Driving Visual Perception, which encompasses
static tasks such as lane detection and drivable area determination, as well as dynamic tasks
involving traffic scenes with vehicles, motorcycles, pedestrians, etc. These technologies help
vehicles understand the road conditions and make decisions for safe navigation.
In this paper, we refer to a driving visual perception model called YOLOP and propose
improvements to enhance its accuracy and computational speed. We made adjustments to two
modules and the selection of activation functions, resulting in improved precision and faster
processing. The proposed method achieves impressive results on the challenging and largescale BDD100K dataset. Compared to the original YOLOP, our approach demonstrates superior
performance across various metrics. In the drivable area detection experiment, we achieved an
accuracy of 97.6%. For lane detection, the accuracy reached 73.6%. In the traffic object
detection experiment, the recall rate reached 88.3%, with an average accuracy of 76.9%.
Furthermore, our approach achieves a speed of 87.7 frames per second (FPS), representing a
51.7% improvement over the original YOLOP.

摘要........................................................................................................................i
Abstract .................................................................................................................ii
誌謝......................................................................................................................iii
目錄......................................................................................................................iv
表目錄..................................................................................................................vi
圖目錄.................................................................................................................vii
第一章緒論...................................................................................................... 1
1.1 研究背景.............................................................................................. 1
1.2 研究動機與目的.................................................................................. 2
1.3 論文架構.............................................................................................. 2
第二章相關研究.............................................................................................. 3
2.1 文獻回顧.............................................................................................. 3
2.2 深度學習與相關技術.......................................................................... 6
第三章改良式 YOLOP ................................................................................. 15
3.1 YOLOP .............................................................................................. 15
3.2 Improved YOLOP.............................................................................. 18
第四章實驗結果與分析................................................................................ 25
4.1 數據集................................................................................................ 25
v
4.2 實驗設定............................................................................................ 29
4.3 實驗結果............................................................................................ 30
第五章結論與未來展望................................................................................ 38
參考文獻............................................................................................................. 39
附件一 實驗環境介紹....................................................................................... 43
附件二 評估指標介紹....................................................................................... 45
Extended Abstract ............................................................................................... 49
Chapter 1 Introduction ........................................................................................ 50
Chapter 2 Related work....................................................................................... 50
Chapter 3 Improved YOLOP .............................................................................. 51
Chapter 4 Experimental Results and Analysis.................................................... 52
Chapter 5 Conclusions ........................................................................................ 53

[1] J. Canny, "A Computational Approach to Edge Detection," IEEE Transactions on
Pattern Analysis and Machine Intelligence, vol. 8, no. 6, pp. 679-698, 1986.
[2] “F. Liu et al., "Low Computation and High Efficiency Sobel Edge Detector for Robot
Vision," 2021 IEEE International Conference on Real-time Computing and Robotics
(RCAR), Xining, China, 2021, pp. 684-689, doi:
10.1109/RCAR52367.2021.9517380.”.
[3] R. O. Duda and P. E. Hart, "Use of the Hough Transformation to Detect Lines and
Curves in Pictures," Comm. ACM, vol. 15, no. 1, pp. 11-15, 1972.
[4] D. H. Ballard, "Generalizing the Hough transform to detect arbitrary shapes," Pattern
Recognition, vol. 13, no. 2, pp. 111-122, 1981.
[5] A. Krizhevsky, I. Sutskever, and G. E. Hinton., "ImageNet Classification with Deep
Convolutional Neural Networks," Advances in Neural Information Processing Systems,
pp. 1097-1105, 2012.
[6] K. Simonyan and A. Zisserman, "Very Deep Convolutional Networks for Large-Scale
Image Recognition," arXiv:1409.1556, 2014. [Online]. Available:
https://arxiv.org/abs/1409.1556.
[7] C. Szegedy, W. Liu, Y. Jia, P. Sermanet, S. Reed, D. Anguelov, D. Erhan, V. Vanhoucke
and A. Rabinovich, "Going deeper with convolutions," in Proceedings of IEEE
Conference on Computer Vision and Pattern Recognition, Boston, MA, USA, 2015.
[8] J. Long, E. Shelhamer, and T. Darrell, "Fully convolutional networks for semantic
segmentation," in Proceedings of IEEE Conference on Computer Vision and Pattern
Recognition, Boston, MA, USA, 2015.
[9] P. Krähenbühl and V. Koltun, "Efficient inference in fully connected CRFs with
Gaussian edge potentials," The 24th International Conference on Neural Information
Processing Systems, pp. 109-117, Oct. 2012.
[10] R. Pascanu, T. Mikolov, and Y. Bengio, "On the Difficulty of Training Recurrent Neural
Networks," in Proceedings of The 30th International Conference on International
Conference on Machine Learning, Atlanta, GA, USA, 2013.
[11] K. He, X. Zhang, S. Ren, and J. Sun, "Deep Residual Learning for Image Recognition,"
in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition,
LasVegas, Nevada, USA, 2016.
[12] J.Zhang, Y.Xu, B.Ni, and Z.Duan, "Geometric constrained joint lane segmentation and
lane boundary detection," in Proceedings of the European Conference on Computer
Vision, Munich, Germany, 2018.
[13] S. Lee, J. Kim, J. S. Yoon, S. Shin, O. Bailo, N. Kim, T. H. Lee, H. S. Hong, S. H. Han,
and I. S. Kweon, "VPGNet: Vanishing point guided network for lane and road marking
detection and recognition," in Proceedings of IEEE International Conference on
Computer Vision, Venice, Italy, 2017.
[14] Gabriel L. Oliveira, Wolfram Burgard and Thomas Brox, “Efficient Deep Models for
Monocular Road Segmentation,” 2016 IEEE/RSJ International Conference on
Intelligent Robots and Systems (IROS), 9-14 October 2016.
[15] Luca Caltagirone, Samuel Scheidegger, Lennart Svensson, Mattias Wahde, “Fast
LIDAR-based Road Detection,” arXiv:1703.03613, 29 MAR 2017.
[16] J. Kim and M. Lee, "Robust Lane Detection Based On Convolutional Neural Network
and Random Sample Consensus," in Proceedings of International Conference on Neural
Information Processing, Cham, Switzerland, 2014.
[17] B. Huval, T. Wang, S. Tandon, J. Kiske, W. Song, J. Pazhayampallil, M. Andriluka, P.
Rajpurkar, T. Migimatsu, R. Cheng-Yue, F. Mujica, A. Coates, and A. Y. Ng, "An
Empirical Evaluation of Deep Learning on Highway Driving," arXiv:1504.01716, 2015.
[Online]. Available: https://arxiv.org/abs/1504.01716.
[18] X. Pan, J. Shi, P. Luo, X. Wang and X. Tang, "Spatial As Deep: Spatial CNN for Traffic
Scene Understanding," arXiv:1712.06080, 2018. [Online]. Available:
https://arxiv.org/abs/1712.06080.
[19] Ze Wang, Weiqiang Ren, Qiang Qiu, “LaneNet: Real-Time Lane Detection Networks
for Autonomous Driving,” arXiv:1807.01726, 4 july 2018.
[20] Lucas Tabelini, Rodrigo Berriel, Thiago M. Paixão, Claudine Badue, Alberto F. De
Souza, Thiago Oliveira-Santos, “Keep your Eyes on the Lane: Real-time Attentionguided Lane Detection,” Computer Vision and Pattern Recognition, 18 nov 2020.
[21] Z. Qin, H. Wang and X. Li, "Ultra Fast Structure-aware Deep Lane Detection,"
arXiv:2004.11757, 2020. [Online]. Available: https://arxiv.org/abs/2004.11757.
[22] Zhengyang Feng,Shaohua Guo, Xin Tan,Ke Xu,Min Wang,Lizhuang Ma, “Rethinking
Efficient Lane Detection via Curve Modeling,” arXiv:2203.02431v1 [cs.CV], 4 mar
2022.
[23] R. Girshick, J. Donahue, T. Darrell, and J. Malik, "Rich feature hierarchies for accurate
object detection and semantic segmentation," in Proceedings of IEEE Conference on
Computer Vision and Pattern Recognition, Columbus, OH, USA, 2014.
[24] R. Girshick, "Fast R-CNN," in Proceedings of IEEE International Conference on
Computer Vision, Santiago, Chile, 2015.
[25] S. Ren, K. He, R. Girshick, and J. Sun, "Faster R-CNN: Towards real-time object
detection with region proposal networks," IEEE Transactions on Pattern Analysis and
Machine Intelligence, vol. 39, no. 6, pp. 1137-1149, Jun 2017.
[26] J. Dai, Y. Li, K. He, and J. Sun, "R-FCN: Object detection via region-based fully
convolutional networks," in Proceeding of the 30th Conference on Neural Information
Processing Systems, Barcelona, Spain, 2016.
[27] J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, "You only look once: unified, realtime object detection," in Proceedings of IEEE Conference on Computer Vision and
Pattern Recognition, Las Vegas, NV, USA, 2016.
[28] W. Liu, D. Anguelov, D. Erhan, C. Szegedy, S. Reed, C.Y. Fu, and A.C. Berg, "SSD:
single shot multibox detector," European Conference on Computer Vision, pp. 21-37,
2016.
[29] T. Y. Lin, P. Goyal, R. Girshick, K. He, and P. Dollar, "Focal loss for dense object
detection," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 42,
no. 2, pp. 318-327, Feb 2020.
[30] Ilya Tolstikhin, Neil Houlsby, Alexander Kolesnikov, Lucas Beyer, Xiaohua Zhai,
Thomas Unterthiner, Jessica Yung, Andreas Steiner, Daniel Keysers, Jakob Uszkoreit,
Mario Lucic, Alexey Dosovitskiy, “MLP-Mixer: An all-MLP Architecture for
Vision,” arXiv:2105.01601 , 11 Jun 2021.
[31] Christian Szegedy, Sergey Ioffe, Vincent Vanhoucke, Alex Alemi, "Inception-v4,
Inception-ResNet and the Impact of Residual Connections on Learning," Computer
Vision and Pattern Recognition, 23 Aug 2016.
[32] K. He, X. Zhang, S. Ren and J. Sun, “Spatial Pyramid Pooling in Deep Convolutional
Networks for Visual Recognition,” in IEEE Transactions on Pattern Analysis and
Machine Intelligence, pp. vol. 37, no. 9, pp. 1904-1916,, 2015.
[33] T. -Y. Lin, P. Dollár, R. Girshick, K. He, B. Hariharan and S. Belongie, "Feature
Pyramid Networks for Object Detection," IEEE Conference on Computer Vision and
Pattern Recognition (CVPR), pp. pp. 936-944, 2017.
[34] C.Wang, A.Bochkovskiy, H. Liao, “Scaled-YOLOv4: Scaling Cross Stage Partial
Network,” Computer Vision and Pattern Recognition, 22 feb 2021.
[35] S. Liu, L. Qi, H. Qin, J. Shi and J. Jia, "Path Aggregation Network for Instance
Segmentation," IEEE/CVF Conference on Computer Vision and Pattern Recognition,
pp. pp. 8759-8768, 2018.
[36] Juan R. Terven,Diana M. Cordova-Esparaza, "A COMPREHENSIVE REVIEW OF
YOLO: FROM YOLOV1 AND BEYOND," arXiv:2304.00501v2 [cs.CV] , 19 May
2023.
[37] C.Yao Wang, A. Bochkovskiy, H. Mark Liao, “YOLOv7: Trainable bag-of-freebies
sets new state-of-the-art for real-time object detectors,” Computer Vision and Pattern
Recognition, 6 jul 2022.
[38] D. Wu, M. Liao, W. Zhang, X. Wang, X. Bai, W.Cheng & W. Liu , “YOLOP: You
Only Look Once for Panoptic Driving Perception,” Machine Intelligence Research,
pp. vol.19, no.6, pp.550–562, 2022.
[39] F. Yu, H. Chen, X. Wang, W. Xian, Y. Chen, F. Liu, V. Madhavan, T. Darrell,
"BDD100K: A Diverse Driving Dataset for Heterogeneous Multitask Learning,"
Published at IEEE Conference on Computer Vision and Pattern Recognition, 12 5 2018.