ShuffleSeg: Real-time Semantic Segmentation Network
This addresses the need for efficient segmentation in mobile/robotics applications, though it appears incremental as it builds on existing techniques like grouped convolution and channel shuffling.
The authors tackled real-time semantic segmentation for mobile and robotics by proposing ShuffleSeg, a computationally efficient network that achieves 2x GFLOPs reduction and 58.3% mIoU on CityScapes while running at 15.7 FPS on NVIDIA Jetson TX2.
Real-time semantic segmentation is of significant importance for mobile and robotics related applications. We propose a computationally efficient segmentation network which we term as ShuffleSeg. The proposed architecture is based on grouped convolution and channel shuffling in its encoder for improving the performance. An ablation study of different decoding methods is compared including Skip architecture, UNet, and Dilation Frontend. Interesting insights on the speed and accuracy tradeoff is discussed. It is shown that skip architecture in the decoding method provides the best compromise for the goal of real-time performance, while it provides adequate accuracy by utilizing higher resolution feature maps for a more accurate segmentation. ShuffleSeg is evaluated on CityScapes and compared against the state of the art real-time segmentation networks. It achieves 2x GFLOPs reduction, while it provides on par mean intersection over union of 58.3% on CityScapes test set. ShuffleSeg runs at 15.7 frames per second on NVIDIA Jetson TX2, which makes it of great potential for real-time applications.