EvoPose2D: Pushing the Boundaries of 2D Human Pose Estimation Using Accelerated Neuroevolution With Weight Transfer
Neural architecture search has proven to be highly effective in the design of efficient convolutional neural networks that are better suited for mobile deployment than hand-designed networks. Hypothesizing that neural architecture search holds great potential for human pose estimation, we explore th...
Main Authors: | , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
IEEE
2021-01-01
|
Series: | IEEE Access |
Subjects: | |
Online Access: | https://ieeexplore.ieee.org/document/9559918/ |
_version_ | 1811274072451973120 |
---|---|
author | William McNally Kanav Vats Alexander Wong John McPhee |
author_facet | William McNally Kanav Vats Alexander Wong John McPhee |
author_sort | William McNally |
collection | DOAJ |
description | Neural architecture search has proven to be highly effective in the design of efficient convolutional neural networks that are better suited for mobile deployment than hand-designed networks. Hypothesizing that neural architecture search holds great potential for human pose estimation, we explore the application of neuroevolution, a form of neural architecture search inspired by biological evolution, in the design of 2D human pose networks for the first time. Additionally, we propose a new weight transfer scheme that enables us to accelerate neuroevolution in a flexible manner. Our method produces network designs that are more efficient and more accurate than state-of-the-art hand-designed networks. In fact, the generated networks process images at higher resolutions using less computation than previous hand-designed networks at lower resolutions, allowing us to push the boundaries of 2D human pose estimation. Our base network designed via neuroevolution, which we refer to as EvoPose2D-S, achieves comparable accuracy to SimpleBaseline while being 50% faster and <inline-formula> <tex-math notation="LaTeX">$12.7\times $ </tex-math></inline-formula> smaller in terms of file size. Our largest network, EvoPose2D-L, achieves new state-of-the-art accuracy on the Microsoft COCO Keypoints benchmark, is <inline-formula> <tex-math notation="LaTeX">$4.3\times $ </tex-math></inline-formula> smaller than its nearest competitor, and has similar inference speed. The code is publicly available at <uri>https://github.com/wmcnally/evopose2d</uri>. |
first_indexed | 2024-04-12T23:12:22Z |
format | Article |
id | doaj.art-0e2a76edec5e4497b8a8ae2b4ec6a057 |
institution | Directory Open Access Journal |
issn | 2169-3536 |
language | English |
last_indexed | 2024-04-12T23:12:22Z |
publishDate | 2021-01-01 |
publisher | IEEE |
record_format | Article |
series | IEEE Access |
spelling | doaj.art-0e2a76edec5e4497b8a8ae2b4ec6a0572022-12-22T03:12:46ZengIEEEIEEE Access2169-35362021-01-01913940313941410.1109/ACCESS.2021.31182079559918EvoPose2D: Pushing the Boundaries of 2D Human Pose Estimation Using Accelerated Neuroevolution With Weight TransferWilliam McNally0https://orcid.org/0000-0002-7187-7147Kanav Vats1Alexander Wong2https://orcid.org/0000-0001-5729-5899John McPhee3https://orcid.org/0000-0003-3908-9519Department of Systems Design Engineering, University of Waterloo, Waterloo, ON, CanadaDepartment of Systems Design Engineering, University of Waterloo, Waterloo, ON, CanadaDepartment of Systems Design Engineering, University of Waterloo, Waterloo, ON, CanadaDepartment of Systems Design Engineering, University of Waterloo, Waterloo, ON, CanadaNeural architecture search has proven to be highly effective in the design of efficient convolutional neural networks that are better suited for mobile deployment than hand-designed networks. Hypothesizing that neural architecture search holds great potential for human pose estimation, we explore the application of neuroevolution, a form of neural architecture search inspired by biological evolution, in the design of 2D human pose networks for the first time. Additionally, we propose a new weight transfer scheme that enables us to accelerate neuroevolution in a flexible manner. Our method produces network designs that are more efficient and more accurate than state-of-the-art hand-designed networks. In fact, the generated networks process images at higher resolutions using less computation than previous hand-designed networks at lower resolutions, allowing us to push the boundaries of 2D human pose estimation. Our base network designed via neuroevolution, which we refer to as EvoPose2D-S, achieves comparable accuracy to SimpleBaseline while being 50% faster and <inline-formula> <tex-math notation="LaTeX">$12.7\times $ </tex-math></inline-formula> smaller in terms of file size. Our largest network, EvoPose2D-L, achieves new state-of-the-art accuracy on the Microsoft COCO Keypoints benchmark, is <inline-formula> <tex-math notation="LaTeX">$4.3\times $ </tex-math></inline-formula> smaller than its nearest competitor, and has similar inference speed. The code is publicly available at <uri>https://github.com/wmcnally/evopose2d</uri>.https://ieeexplore.ieee.org/document/9559918/Artificial intelligencecomputer visionconvolutional neural networkdeep learninghuman pose estimationneural architecture search |
spellingShingle | William McNally Kanav Vats Alexander Wong John McPhee EvoPose2D: Pushing the Boundaries of 2D Human Pose Estimation Using Accelerated Neuroevolution With Weight Transfer IEEE Access Artificial intelligence computer vision convolutional neural network deep learning human pose estimation neural architecture search |
title | EvoPose2D: Pushing the Boundaries of 2D Human Pose Estimation Using Accelerated Neuroevolution With Weight Transfer |
title_full | EvoPose2D: Pushing the Boundaries of 2D Human Pose Estimation Using Accelerated Neuroevolution With Weight Transfer |
title_fullStr | EvoPose2D: Pushing the Boundaries of 2D Human Pose Estimation Using Accelerated Neuroevolution With Weight Transfer |
title_full_unstemmed | EvoPose2D: Pushing the Boundaries of 2D Human Pose Estimation Using Accelerated Neuroevolution With Weight Transfer |
title_short | EvoPose2D: Pushing the Boundaries of 2D Human Pose Estimation Using Accelerated Neuroevolution With Weight Transfer |
title_sort | evopose2d pushing the boundaries of 2d human pose estimation using accelerated neuroevolution with weight transfer |
topic | Artificial intelligence computer vision convolutional neural network deep learning human pose estimation neural architecture search |
url | https://ieeexplore.ieee.org/document/9559918/ |
work_keys_str_mv | AT williammcnally evopose2dpushingtheboundariesof2dhumanposeestimationusingacceleratedneuroevolutionwithweighttransfer AT kanavvats evopose2dpushingtheboundariesof2dhumanposeestimationusingacceleratedneuroevolutionwithweighttransfer AT alexanderwong evopose2dpushingtheboundariesof2dhumanposeestimationusingacceleratedneuroevolutionwithweighttransfer AT johnmcphee evopose2dpushingtheboundariesof2dhumanposeestimationusingacceleratedneuroevolutionwithweighttransfer |