ScaleNAS: Multi-Path One-Shot NAS for Scale-Aware High-Resolution Representation

Hsin-Pai Cheng, Feng Liang, Meng Li, Bowen Cheng, Feng Yan, Hai Li, Vikas Chandra, Yiran Chen
Proceedings of the First International Conference on Automated Machine Learning, PMLR 188:15/1-18, 2022.

Abstract

Scale variance among different sizes of body parts and objects is a challenging problem for visual recognition tasks. Existing works usually design dedicated backbone or apply Neural architecture Search (NAS) for each task to tackle this challenge. However, existing works impose significant limitations on the design or search space. To solve these problems, we present ScaleNAS, a one-shot learning method for exploring scale-aware representations. ScaleNAS solves multiple tasks at a time by searching multi-scale feature aggregation. ScaleNAS adopts a flexible search space that allows an arbitrary number of blocks and cross-scale feature fusions. To cope with the high search cost incurred by the flexible space, ScaleNAS employs one-shot learning for multi-scale supernet driven by grouped sampling and evolutionary search. Without further retraining, ScaleNet can be directly deployed for different visual recognition tasks with superior performance. We use ScaleNAS to create high-resolution models for two different tasks, ScaleNet-P for human pose estimation and ScaleNet-S for semantic segmentation. ScaleNet-P and ScaleNet-S outperform existing manually crafted and NAS-based methods in both tasks. When applying ScaleNet-P to bottom-up human pose estimation, it surpasses the state-of-the-art HigherHRNet. In particular, ScaleNet-P4 achieves 71.6% AP on COCO test-dev, achieving new state-of-the-art result.

Cite this Paper

BibTeX
@InProceedings{pmlr-v188-cheng22a, title = {ScaleNAS: Multi-Path One-Shot NAS for Scale-Aware High-Resolution Representation}, author = {Cheng, Hsin-Pai and Liang, Feng and Li, Meng and Cheng, Bowen and Yan, Feng and Li, Hai and Chandra, Vikas and Chen, Yiran}, booktitle = {Proceedings of the First International Conference on Automated Machine Learning}, pages = {15/1--18}, year = {2022}, editor = {Guyon, Isabelle and Lindauer, Marius and van der Schaar, Mihaela and Hutter, Frank and Garnett, Roman}, volume = {188}, series = {Proceedings of Machine Learning Research}, month = {25--27 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v188/cheng22a/cheng22a.pdf}, url = {https://proceedings.mlr.press/v188/cheng22a.html}, abstract = {Scale variance among different sizes of body parts and objects is a challenging problem for visual recognition tasks. Existing works usually design dedicated backbone or apply Neural architecture Search (NAS) for each task to tackle this challenge. However, existing works impose significant limitations on the design or search space. To solve these problems, we present ScaleNAS, a one-shot learning method for exploring scale-aware representations. ScaleNAS solves multiple tasks at a time by searching multi-scale feature aggregation. ScaleNAS adopts a flexible search space that allows an arbitrary number of blocks and cross-scale feature fusions. To cope with the high search cost incurred by the flexible space, ScaleNAS employs one-shot learning for multi-scale supernet driven by grouped sampling and evolutionary search. Without further retraining, ScaleNet can be directly deployed for different visual recognition tasks with superior performance. We use ScaleNAS to create high-resolution models for two different tasks, ScaleNet-P for human pose estimation and ScaleNet-S for semantic segmentation. ScaleNet-P and ScaleNet-S outperform existing manually crafted and NAS-based methods in both tasks. When applying ScaleNet-P to bottom-up human pose estimation, it surpasses the state-of-the-art HigherHRNet. In particular, ScaleNet-P4 achieves 71.6% AP on COCO test-dev, achieving new state-of-the-art result.} }
Endnote
%0 Conference Paper %T ScaleNAS: Multi-Path One-Shot NAS for Scale-Aware High-Resolution Representation %A Hsin-Pai Cheng %A Feng Liang %A Meng Li %A Bowen Cheng %A Feng Yan %A Hai Li %A Vikas Chandra %A Yiran Chen %B Proceedings of the First International Conference on Automated Machine Learning %C Proceedings of Machine Learning Research %D 2022 %E Isabelle Guyon %E Marius Lindauer %E Mihaela van der Schaar %E Frank Hutter %E Roman Garnett %F pmlr-v188-cheng22a %I PMLR %P 15/1--18 %U https://proceedings.mlr.press/v188/cheng22a.html %V 188 %X Scale variance among different sizes of body parts and objects is a challenging problem for visual recognition tasks. Existing works usually design dedicated backbone or apply Neural architecture Search (NAS) for each task to tackle this challenge. However, existing works impose significant limitations on the design or search space. To solve these problems, we present ScaleNAS, a one-shot learning method for exploring scale-aware representations. ScaleNAS solves multiple tasks at a time by searching multi-scale feature aggregation. ScaleNAS adopts a flexible search space that allows an arbitrary number of blocks and cross-scale feature fusions. To cope with the high search cost incurred by the flexible space, ScaleNAS employs one-shot learning for multi-scale supernet driven by grouped sampling and evolutionary search. Without further retraining, ScaleNet can be directly deployed for different visual recognition tasks with superior performance. We use ScaleNAS to create high-resolution models for two different tasks, ScaleNet-P for human pose estimation and ScaleNet-S for semantic segmentation. ScaleNet-P and ScaleNet-S outperform existing manually crafted and NAS-based methods in both tasks. When applying ScaleNet-P to bottom-up human pose estimation, it surpasses the state-of-the-art HigherHRNet. In particular, ScaleNet-P4 achieves 71.6% AP on COCO test-dev, achieving new state-of-the-art result.
APA
Cheng, H., Liang, F., Li, M., Cheng, B., Yan, F., Li, H., Chandra, V. & Chen, Y.. (2022). ScaleNAS: Multi-Path One-Shot NAS for Scale-Aware High-Resolution Representation. Proceedings of the First International Conference on Automated Machine Learning, in Proceedings of Machine Learning Research 188:15/1-18 Available from https://proceedings.mlr.press/v188/cheng22a.html.

Related Material