Fast-SCNN: Fast Semantic Segmentation Network

Code

• github.com/open-mmlab/mmsegmentation
  pytorch★ 9,690
• github.com/PaddlePaddle/PaddleSeg
  paddle★ 9,319
• github.com/osmr/imgclsmob
  mxnet★ 3,015
• github.com/zh320/realtime-semantic-segmentation-pytorch
  pytorch★ 249
• github.com/SkyWa7ch3r/ImageSegmentation
  tf★ 7
• github.com/Tramac/Fast-SCNN-pytorch
  pytorch★ 0
• github.com/rachthree/fast_scnn
  tf★ 0
• github.com/SkyWa7ch3r/SceneSegmentation
  tf★ 0
• github.com/kshitizrimal/Fast-SCNN
  tf★ 0
• github.com/MindSpore-paper-code-3/code3/tree/main/fastscnn
  mindspore★ 0

Abstract

The encoder-decoder framework is state-of-the-art for offline semantic image segmentation. Since the rise in autonomous systems, real-time computation is increasingly desirable. In this paper, we introduce fast segmentation convolutional neural network (Fast-SCNN), an above real-time semantic segmentation model on high resolution image data (1024x2048px) suited to efficient computation on embedded devices with low memory. Building on existing two-branch methods for fast segmentation, we introduce our `learning to downsample' module which computes low-level features for multiple resolution branches simultaneously. Our network combines spatial detail at high resolution with deep features extracted at lower resolution, yielding an accuracy of 68.0% mean intersection over union at 123.5 frames per second on Cityscapes. We also show that large scale pre-training is unnecessary. We thoroughly validate our metric in experiments with ImageNet pre-training and the coarse labeled data of Cityscapes. Finally, we show even faster computation with competitive results on subsampled inputs, without any network modifications.

Tasks

Benchmark Results

Reproductions