Method Details

Details for method 'FasterSeg'


Method overview

name FasterSeg
challenge pixel-level semantic labeling
details We present FasterSeg, an automatically designed semantic segmentation network with not only state-of-the-art performance but also faster speed than current methods. Utilizing neural architecture search (NAS), FasterSeg is discovered from a novel and broader search space integrating multi-resolution branches, that has been recently found to be vital in manually designed segmentation models. To better calibrate the balance between the goals of high accuracy and low latency, we propose a decoupled and fine-grained latency regularization, that effectively overcomes our observed phenomenons that the searched networks are prone to "collapsing" to low-latency yet poor-accuracy models. Moreover, we seamlessly extend FasterSeg to a new collaborative search (co-searching) framework, simultaneously searching for a teacher and a student network in the same single run. The teacher-student distillation further boosts the student model's accuracy. Experiments on popular segmentation benchmarks demonstrate the competency of FasterSeg. For example, FasterSeg can run over 30% faster than the closest manually designed competitor on Cityscapes, while maintaining comparable accuracy.
publication FasterSeg: Searching for Faster Real-time Semantic Segmentation
Wuyang Chen, Xinyu Gong, Xianming Liu, Qian Zhang, Yuan Li, Zhangyang Wang
ICLR 2020
project page / code
used Cityscapes data fine annotations
used external data
runtime 0.00613 s
1080Ti, TensorRT v5.1.5
subsampling no
submission date November, 2019
previous submissions


Average results

Metric Value
IoU Classes 71.4577
iIoU Classes 44.3255
IoU Categories 88.0913
iIoU Categories 73.5731


Class results

Class IoU iIoU
road 98.0238 -
sidewalk 83.4809 -
building 91.1373 -
wall 39.1329 -
fence 48.718 -
pole 58.6123 -
traffic light 66.7319 -
traffic sign 71.6239 -
vegetation 92.3337 -
terrain 69.119 -
sky 94.4914 -
person 81.4742 61.0181
rider 61.7829 37.851
car 93.6808 87.6284
truck 54.9549 21.0187
bus 67.0913 34.2727
train 61.1194 28.4967
motorcycle 54.9739 30.4605
bicycle 69.2144 53.8577


Category results

Category IoU iIoU
flat 98.2413 -
nature 92.0576 -
object 65.59 -
sky 94.4914 -
construction 91.6315 -
human 81.7991 62.247
vehicle 92.828 84.8992



Download results as .csv file

Benchmark page