Category-Level Pose Retrieval with Contrastive Features Learnt with Occlusion Augmentation

TL;DR

This paper introduces a category-level pose retrieval method using contrastive learning with occlusion augmentation, achieving state-of-the-art results efficiently without direct pose prediction.

Contribution

It proposes a contrastive embedding approach with a dynamic margin for category-level pose estimation, improving robustness and efficiency over traditional render-and-compare methods.

Findings

Achieves state-of-the-art on PASCAL3D and OccludedPASCAL3D datasets.

Surpasses existing methods on KITTI3D in cross-dataset evaluation.

Provides a real-time image retrieval system for pose estimation.

Abstract

Pose estimation is usually tackled as either a bin classification or a regression problem. In both cases, the idea is to directly predict the pose of an object. This is a non-trivial task due to appearance variations between similar poses and similarities between dissimilar poses. Instead, we follow the key idea that comparing two poses is easier than directly predicting one. Render-and-compare approaches have been employed to that end, however, they tend to be unstable, computationally expensive, and slow for real-time applications. We propose doing category-level pose estimation by learning an alignment metric in an embedding space using a contrastive loss with a dynamic margin and a continuous pose-label space. For efficient inference, we use a simple real-time image retrieval scheme with a pre-rendered and pre-embedded reference set of renderings. To achieve robustness to real-world conditions, we employ synthetic occlusions, bounding box perturbations, and appearance augmentations. Our approach achieves state-of-the-art performance on PASCAL3D and OccludedPASCAL3D and surpasses the competing methods on KITTI3D in a cross-dataset evaluation setting. The code is currently available at https://github.com/gkouros/contrastive-pose-retrieval.