Application of R-FCN Algorithm in Machine Visual Solutions on Tensorflow Based
This paper presents a solution based on Tensorflow platform and R - FCN deep learning model about self-driving cars image processing. Through the Supervised learning of data sets, make them exercise the image segmentation and recognition of information, thus to self-driving cars driving decision-making support.
KeywordsDeep learning Image processing Machine vision Autonomous driving
Thanks to Beimen Shenzhou Special Vehicle Laboratory, School of Computer Science, Beijing Information Science and Technology University, School of Vehicle Engineering, Tsinghua University.
- 1.TensorFlow. https://en.wikipedia.org/wiki/TensorFlow
- 2.Nan, Y.: Research on Convolutional Neural Networks Based on Caffe Deep Learning Framework. Hebei Normal University, Hebei (2014)Google Scholar
- 3.Xiaochun, L.: Research on Classification and Recognition of Handwritten Images Based on Deep Learning. Donghua Institute of Measurement and Computing, Nanchang (2016)Google Scholar
- 4.Bouvrie, J.: Notes on Convolutional Neural Networks (2006)Google Scholar
- 5.Fully Convolutional Networks for Semantic Segmentation, CVPR 2015 best paper, key word: pixel level, fully supervised, CNN (2015)Google Scholar
- 6.Jifeng, D.: Object Detection via Region-Based Fully Convolutional Networks. People’s Posts and Telecommunications Press (2016)Google Scholar
- 7.Girshick, R.: Fast R-CNN 2016 (2015)Google Scholar
- 8.Fu, R.: Research on Image Target Recognition Based on Deep Learning. National University of Defense Technology, Changsha (2014)Google Scholar
- 9.Guiying, Z.: Review of Research on Autopilot Algorithm Based on Computer Vision. Guizhou University, Guizhou (2016)Google Scholar