Method Details

Details for method 'Auto-DeepLab-L'


Method overview

name Auto-DeepLab-L
challenge pixel-level semantic labeling
details In this work, we study Neural Architecture Search for semantic image segmentation, an important computer vision task that assigns a semantic label to every pixel in an image. Existing works often focus on searching the repeatable cell structure, while hand-designing the outer network structure that controls the spatial resolution changes. This choice simplifies the search space, but becomes increasingly problematic for dense image prediction which exhibits a lot more network level architectural variations. Therefore, we propose to search the network level structure in addition to the cell level structure, which forms a hierarchical architecture search space. We present a network level search space that includes many popular designs, and develop a formulation that allows efficient gradient-based architecture search (3 P100 GPU days on Cityscapes images). We demonstrate the effectiveness of the proposed method on the challenging Cityscapes, PASCAL VOC 2012, and ADE20K datasets. Without any ImageNet pretraining, our architecture searched specifically for semantic image segmentation attains state-of-the-art performance. Please refer to for details.
publication Auto-DeepLab: Hierarchical Neural Architecture Search for Semantic Image Segmentation
Chenxi Liu, Liang-Chieh Chen, Florian Schroff, Hartwig Adam, Wei Hua, Alan Yuille, Li Fei-Fei
project page / code
used Cityscapes data fine annotations, coarse annotations
used external data
runtime n/a
subsampling no
submission date January, 2019
previous submissions


Average results

Metric Value
IoU Classes 82.0877
iIoU Classes 61.0204
IoU Categories 91.8881
iIoU Categories 82.0288


Class results

Class IoU iIoU
road 98.8052 -
sidewalk 87.5807 -
building 93.8336 -
wall 61.4061 -
fence 64.4019 -
pole 71.229 -
traffic light 77.6402 -
traffic sign 80.9249 -
vegetation 94.0517 -
terrain 72.742 -
sky 96.0189 -
person 87.7659 73.3552
rider 72.7636 51.3271
car 96.4765 91.6261
truck 78.2179 43.8483
bus 90.8601 56.8466
train 88.4173 55.7123
motorcycle 68.9735 50.1033
bicycle 77.5568 65.3446


Category results

Category IoU iIoU
flat 98.7916 -
nature 93.7239 -
object 76.659 -
sky 96.0189 -
construction 94.1629 -
human 87.8364 74.2399
vehicle 96.0237 89.8177



Download results as .csv file

Benchmark page