MV-ROPE: Multi-view Constraints for Robust Category-level Object Pose and Size Estimation

TL;DR

This paper introduces MV-ROPE, a multi-view framework that leverages RGB video streams, SLAM, and pose graph optimization to estimate object pose and size accurately without relying on high-quality depth sensors.

Contribution

The paper presents a novel multi-view approach combining SLAM, a lightweight pose predictor, and pose graph optimization for robust category-level object pose and size estimation from RGB videos.

Findings

Achieves comparable accuracy to RGB-D methods with high-quality depth data.

Outperforms previous RGB-based methods in scenarios with limited or no depth information.

Demonstrates robustness and accuracy across multiple datasets with varying depth quality.

Abstract

Recently there has been a growing interest in category-level object pose and size estimation, and prevailing methods commonly rely on single view RGB-D images. However, one disadvantage of such methods is that they require accurate depth maps which cannot be produced by consumer-grade sensors. Furthermore, many practical real-world situations involve a moving camera that continuously observes its surroundings, and the temporal information of the input video streams is simply overlooked by single-view methods. We propose a novel solution that makes use of RGB video streams. Our framework consists of three modules: a scale-aware monocular dense SLAM solution, a lightweight object pose predictor, and an object-level pose graph optimizer. The SLAM module utilizes a video stream and additional scale-sensitive readings to estimate camera poses and metric depth. The object pose predictor then generates canonical object representations from RGB images. The object pose is estimated through geometric registration of these canonical object representations with estimated object depth points. All per-view estimates finally undergo optimization within a pose graph, culminating in the output of robust and accurate canonical object poses. Our experimental results demonstrate that when utilizing public dataset sequences with high-quality depth information, the proposed method exhibits comparable performance to state-of-the-art RGB-D methods. We also collect and evaluate on new datasets containing depth maps of varying quality to further quantitatively benchmark the proposed method alongside previous RGB-D based methods. We demonstrate a significant advantage in scenarios where depth input is absent or the quality of depth sensing is limited.