GPU-Efficient Dense Convolutional Network for Real-time Semantic Segmentation

Real-time semantic segmentation is a challenging task as both accuracy and inference speed need to be considered simultaneously. In real-world applications, it is usually achieved by deploying a deep neural network in modern GPU device. However, most of the work focused on real-time semantic segmentation is designed by significantly reducing computation complexity and model size. There are other factors that have a significant impact on inference speed are overlooked, especially when the network is running in modern GPU device. In this paper, we focus on designing a GPU-efficient network as backbone for real-time semantic segmentation. Dense connectivity can preserve and accumulate feature maps of multiple receptive fields and is therefore ideal for semantic segmentation. Therefore, we design a GPU-efficient network (DenseENet) with dense connectivity. The proposed DenseENet shows an obvious advantage in balancing accuracy and inference speed in modern GPU device. Specifically, on Cityscapes test set, DenseENet with a simple FCN decoder achieves 75. 2% mIoU with 83. 6 FPS for an input of 1024 × 2048 resolution and 73. 6% mIoU with 132 FPS for an input of 768 × 1536 resolution on a single GTX 1080Ti card.

Authors

• Xinneng Yang
• Yan Wu 0011
• Junqiao Zhao
• Feilin Liu

Keywords

• Deep learning
• Automation
• Conferences
• Computational modeling
• Semantics
• Graphics processing units
• Real-time systems
• Convolutional Network
• Semantic Segmentation
• Semantic Segmentation Network
• Real-time Semantic Segmentation
• Deep Neural Network
• Feature Maps
• Model Size
• Receptive Field
• Dense Connections
• Fully Convolutional Network
• Inference Speed
• Modern Devices
• Growth Rate
• Computational Cost
• Factorization
• Input Features
• Dense Layer
• Convolution Operation
• Bottleneck Layer
• Pointwise Convolution
• Convolutional Neural Network Model
• Output Feature Map
• Neural Architecture Search
• Degree Of Parallelism
• Number Of Input Features
• Transition Layer
• Batch Normalization
• Kernel Parameters