This article explains the method of optimizing the object detection model used in the automatic drive. At present, there are a lot of car accidents happen caused by the low robustness of the automatic pilot. We are focused on improving the robustness of the model in a unique environment. We built four special environments and collected the images of the environment in a special way, and used them to train the detectron2, which is our initial model. Adversarial training is the way to train the model. Finally, we optimize the model successfully. If more people are willing to spend time and money training a particular model for a location, the automatic pilot can be promoted to any place.

Ningchan Wang. Robust Object Detection Model for Autonomous Driving in Real Scenarios. Academic Journal of Computing & Information Science (2023), Vol. 6, Issue 3: 52-61. https://doi.org/10.25236/AJCIS.2023.060307.

[1] Tian Yuchi, Kexin Pei, Suman Jana, and Baishakhi Ray (2018). "Deeptest: Automated testing of deep-neural-network-driven autonomous cars." In Proceedings of the 40th international conference on software engineering, pp303-314.

[2] Zhao Zhongqiu, Peng Zheng, Shoutao Xu, and Xindong Wu (2019). "Object detection with deep learning: A review." IEEE transactions on neural networks and learning systems 30, no. 11: pp3212-3232.

[3] Slack, Dylan, Sophie Hilgard, Emily Jia, Sameer Singh, and Himabindu Lakkaraju(2020). "Fooling lime and shap: Adversarial attacks on post hoc explanation methods." In Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society, pp. 180-186.

[4] George Karimpanal, Thommen; Bouffanais, Roland (2019). "Self-organizing maps for storage and transfer of knowledge in reinforcement learning". Adaptive Behavior. 27 (2):pp111–126.

[5] Tsung-Yi Lin, Piotr Dollar, Ross Girshick, Kaiming He, Bharath Hariharan(2017). Serge Belongie; Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 2117-2125.