AutoPose: Searching Multi-Scale Branch Aggregation for Pose Estimation

Abstract

We present AutoPose, a novel neural architecture search(NAS) framework that is capable of automatically discovering multiple parallel branches of cross-scale connections towards accurate and high-resolution 2D human pose estimation. Recently, high-performance hand-crafted convolutional networks for pose estimation show growing demands on multi-scale fusion and high-resolution representations. However, current NAS works exhibit limited flexibility on scale searching, they dominantly adopt simplified search spaces of single-branch architectures. Such simplification limits the fusion of information at different scales and fails to maintain high-resolution representations. The presentedAutoPose framework is able to search for multi-branch scales and network depth, in addition to the cell-level microstructure. Motivated by the search space, a novel bi-level optimization method is presented, where the network-level architecture is searched via reinforcement learning, and the cell-level search is conducted by the gradient-based method. Within 2.5 GPU days, AutoPose is able to find very competitive architectures on the MS COCO dataset, that are also transferable to the MPII dataset. Our code is available at https://github.com/VITA-Group/AutoPose.

References in corpus (9)

• Neural Architecture Search with Reinforcement Learning
• DARTS: Differentiable Architecture Search
• Neural Architecture Optimization
• Searching for Efficient Multi-Scale Architectures for Dense Image Prediction
• SNAS: Stochastic Neural Architecture Search
• DeepArchitect: Automatically Designing and Training Deep Architectures
• Auto-ReID: Searching for a Part-aware ConvNet for Person Re-Identification
• AutoSpeech: Neural Architecture Search for Speaker Recognition
• Pose Neural Fabrics Search

Cited by in corpus (7)

• Deep Learning-Based Human Pose Estimation: A Survey
• ZoomNAS: Searching for Whole-body Human Pose Estimation in the Wild
• EvoPose2D: Pushing the Boundaries of 2D Human Pose Estimation using Accelerated Neuroevolution with Weight Transfer
• Learning Versatile Neural Architectures by Propagating Network Codes
• ScaleNAS: One-Shot Learning of Scale-Aware Representations for Visual Recognition
• ViPNAS: Efficient Video Pose Estimation via Neural Architecture Search
• HR-NAS: Searching Efficient High-Resolution Neural Architectures with Lightweight Transformers