Physics

Channel (broadcasting)

Block (permutation group theory)

Computer network

Mathematics

Operating system

Algorithm

Optics

Computer engineering

Encoder

Pyramid (geometry)

Residual

Artificial intelligence

Management

Economics

Computational complexity theory

Geometry

Computer science

Computation

Task (project management)

Real-time computing

Computer vision

Decoding methods

Segmentation

The extensive computational burden limits the usage of CNNs in mobile devices for dense estimation tasks. In this paper, we present a lightweight network to address this problem, namely LEDNet, which employs an asymmetric encoder-decoder architecture for the task of real-time semantic segmentation. More specifically, the encoder adopts a ResNet as backbone network, where two new operations, channel split and shuffle, are utilized in each residual block to greatly reduce computation cost while maintaining higher segmentation accuracy. On the other hand, an attention pyramid network (APN) is employed in the decoder to further lighten the entire network complexity. Our model has less than 1M parameters, and is able to run at over 71 FPS in a single GTX 1080Ti GPU. The comprehensive experiments demonstrate that our approach achieves state-of-the-art results in terms of speed and accuracy trade-off on CityScapes dataset.

Lednet: A Lightweight Encoder-Decoder Network for Real-Time Semantic Segmentation