Convolutional Pose Machines

TL;DR

This paper introduces a convolutional pose machine framework that models long-range dependencies for pose estimation, achieving state-of-the-art results without explicit graphical models.

Contribution

It presents a systematic way to incorporate convolutional networks into pose machines, enabling implicit modeling of spatial dependencies and refined part localization.

Findings

Achieves state-of-the-art performance on MPII, LSP, and FLIC datasets.

Addresses vanishing gradient issues with intermediate supervision.

Outperforms existing methods in pose estimation benchmarks.

Abstract

Pose Machines provide a sequential prediction framework for learning rich implicit spatial models. In this work we show a systematic design for how convolutional networks can be incorporated into the pose machine framework for learning image features and image-dependent spatial models for the task of pose estimation. The contribution of this paper is to implicitly model long-range dependencies between variables in structured prediction tasks such as articulated pose estimation. We achieve this by designing a sequential architecture composed of convolutional networks that directly operate on belief maps from previous stages, producing increasingly refined estimates for part locations, without the need for explicit graphical model-style inference. Our approach addresses the characteristic difficulty of vanishing gradients during training by providing a natural learning objective function that enforces intermediate supervision, thereby replenishing back-propagated gradients and conditioning the learning procedure. We demonstrate state-of-the-art performance and outperform competing methods on standard benchmarks including the MPII, LSP, and FLIC datasets.