A real time semantic segmentation method based on multi-level feature fusion

doi:10.13474/j.cnki.11-2246.2020.0003

Abstract

Abstract: Road scene understanding is an important module in autonomous driving as it can provide rich information for other modules like high-res map and dynamic planning. Semantic segmentation classifies each pixel in an image and it is often used in scene understanding. However, common segmentation algorithms cannot balance the accuracy and speed. In this paper, we first extract features with light-weight backbone MobileNetV2, and then introduce a new feature fusion module to combine multi-scale features efficiently, which can makes a good balance between accuracy and speed. Finally, we evaluate our method on the Cityscapes dataset.

Key words: autonomous driving, semantic segmentation, scene understanding, deep learning, multi-scale feature fusion

CLC Number:

P228

ZHOU Jimiao, LI Bijun, CHEN Shizeng. A real time semantic segmentation method based on multi-level feature fusion[J]. Bulletin of Surveying and Mapping, 2020, 0(1): 10-15.

References

[1] 李琳辉, 钱波, 连静, 等. 基于卷积神经网络的交通场景语义分割方法研究[J]. 通信学报, 2018, 39(4):123-130.
[2] KRIZHEVSKY A, SUTSKEVER I, HINTON G E. Image net classification with deep convolutional neural networks[C]//Advances in neural information processing systems. Lake Tahoe:MITPress, 2012:1097-1105.
[3] LECUN Y, BOTTOU L, BENGIO Y, et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE, 1998, 86(11):2278-2324.
[4] LONG J, SHELHAMER E, DARRELL T. Fully convolu-tional networks for semantic segmentation[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. Boston:IEEE, 2015:3431-3440.
[5] CHEN L C, ZHU Y, PAPANDREOU G, et al. Encoder-decoder with atrous separable convolution for semantic image segmentation[C]//Proceedings of the European conference on computer vision (ECCV). Munich:Springer, 2018:801-818.
[6] YU F, KOLTUN V. Multi-Scale context aggregation by dilated convolutions[J/OL]. 2016-04-30[2019-08-21].https://arxiv.org/abs/1511.07122.
[7] ZHAO H, SHI J, QI X, et al. Pyramid scene parsing network[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. HonoluluI:IEEE, 2017:2881-2890.
[8] PASZKE A, CHAURASIA A, KIM S, et al. Enet:a deep neural network architecture for real-time semantic segmentation[J/OL]. 2016-06-07[2019-08-21].https://arxiv.org/abs/1606.02147.
[9] MEHTA S, RASTEGARI M, CASPI A, et al. Espnet:efficient spatial pyramid of dilated convolutions for semantic segmentation[C]//Proceedings of the European Conference on Computer Vision (ECCV). Munich:Springer, 2018:552-568.
[10] SANDLER M, HOWARD A, ZHU M, et al. Mobilenetv2:inverted residuals and linear bottlenecks[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Salt Lake City:IEEE, 2018:4510-4520.
[11] CORDTS M, OMRAN M, RAMOS S, et al. The cityscapes dataset for semantic urban scene understanding[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. Las Vegas:IEEE, 2016:3213-3223.
[12] CHOLLET F. Xception:deep learning with depthwise separable convolutions[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. Honolulu:IEEE, 2017:1251-1258.
[13] LIN M, CHEN Q, YAN S. Network in network[J/OL].2014-03-04[2019-08-22].https://arxiv.org/abs/1312.4400.
[14] ZEILER M D, FERGUS R. Visualizing and understanding convolutional network[C]//European conference on computer vision. Zurich:Springer, 2014:818-833.
[15] PENG C, ZHANG X, YU G, et al. Large kernel matters-improve semantic segmentation by global convolutional network[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. Honolulu:IEEE, 2017:4353-4361.
[16] YU C, WANG J, PENG C, et al. Bisenet:bilateral segmentation network for real-time semantic segmentation[C]//Proceedings of the European Conference on Computer Vision (ECCV). Munich:Springer, 2018:325-341.
[17] BADRINARAYANAN V, KENDALL A, CIPOLLA R. Segnet:a deep convolutional encoder-decoder architecture for image segmentation[J]. IEEE transactions on pattern analysis and machine intelligence, 2017, 39(12):2481-2495.