Efficient Dense Modules of Asymmetric Convolution for Real-Time Semantic Segmentation

[guanfuchen]

related paper

摘要

Real-time semantic segmentation plays an important role in practical applications such as self-driving and robots. Most research working on semantic segmentation focuses on accuracy with little consideration for efficiency. Several existing studies that emphasize high-speed inference often cannot produce high-accuracy segmentation results. In this paper, we propose a novel convolutional network named Efficient Dense modules with Asymmetric convolution (EDANet), which employs an asymmetric convolution structure incorporating the dilated convolution and the dense connectivity to attain high efficiency at low computational cost, inference time, and model size. Compared to FCN, EDANet is 11 times faster and has 196 times fewer parameters, while it achieves a higher the mean of intersection-over-union (mIoU) score without any additional decoder structure, context module, post-processing scheme, and pretrained model. We evaluate EDANet on Cityscapes and CamVid datasets to evaluate its performance and compare it with the other state-of-art systems. Our network can run on resolution 512×1024 inputs at the speed of 108 and 81 frames per second on a single GTX 1080Ti and Titan X, respectively.

[guanfuchen]

Inference speed and mIoU accuracy on Cityscapes test set (Cordts et al. 2016). Networks included are FCN (Long, Shelhamer, and Darrell 2015), DeepLab (Chen et al. 2016), PSPNet (Zhao et al. 2017), SegModel (Shen et al. 2017), Dilation10 (Yu and Koltun 2016), CRF-RNN (Zheng et al. 2015), SegNet (Badrinarayanan, Kendall, and Cipolla 2015), SQ (Treml et al. 2016), ENet (Paszke et al. 2016), ERFNet (Romera et al. 2017), ContextNet (Poudel et al. 2018), SkipNet-MobileNet (Siam et al. 2018), ESPNet (Mehta et al. 2018), and our EDANet.