Deformable Part Networks

TL;DR

Deformable Part Networks (DPNs) are a novel approach for pose-invariant 2D object recognition, modeling deformable parts hierarchically and outperforming existing pose-aware networks on affNIST.

Contribution

The paper introduces DPNs, a new network architecture that models deformable parts hierarchically for pose-invariant recognition, with improved performance over CapsNet and STN.

Findings

DPNs outperform CapsNet and STN on affNIST.

DPNs exhibit better generalization to affine transformations.

A 17-layer DPN achieves significant accuracy improvements.

Abstract

In this paper we propose novel Deformable Part Networks (DPNs) to learn {\em pose-invariant} representations for 2D object recognition. In contrast to the state-of-the-art pose-aware networks such as CapsNet \cite{sabour2017dynamic} and STN \cite{jaderberg2015spatial}, DPNs can be naturally {\em interpreted} as an efficient solver for a challenging detection problem, namely Localized Deformable Part Models (LDPMs) where localization is introduced to DPMs as another latent variable for searching for the best poses of objects over all pixels and (predefined) scales. In particular we construct DPNs as sequences of such LDPM units to model the semantic and spatial relations among the deformable parts as hierarchical composition and spatial parsing trees. Empirically our 17-layer DPN can outperform both CapsNets and STNs significantly on affNIST \cite{sabour2017dynamic}, for instance, by 19.19\% and 12.75\%, respectively, with better generalization and better tolerance to affine transformations.