サイクル一貫性アンカーポイントを活用した自己教師ありRGB-D位置合わせ
Leveraging Cycle-Consistent Anchor Points for Self-Supervised RGB-D Registration
サイクル一貫性のあるキーポイントをアンカーとして用い、GRUと変換同期を組み合わせた新たな姿勢推定ブロックにより、自己教師ありRGB-D位置合わせの精度を向上させた研究。
詳しい要約
1. どんなもの？
2. 先行研究と比べてどこがすごい？
3. 技術・手法の肝は？
4. どうやって有効だと検証した？
5. 議論はある？
6. 次に読むべき論文は？
※ AIが要旨から生成した要約です。正確性は原文をご確認ください。
著者: Siddharth Tourani, Jayaram Reddy, Sarvesh Thakur, K Madhava Krishna, Muhammad Haris Khan, N Dinesh Reddy
分類: cs.CV, cs.RO
原文アブストラクト
With the rise in consumer depth cameras, a wealth of unlabeled RGB-D data has become available. This prompts the question of how to utilize this data for geometric reasoning of scenes. While many RGB-D registration meth- ods rely on geometric and feature-based similarity, we take a different approach. We use cycle-consistent keypoints as salient points to enforce spatial coherence constraints during matching, improving correspondence accuracy. Additionally, we introduce a novel pose block that combines a GRU recurrent unit with transformation synchronization, blending historical and multi-view data. Our approach surpasses previous self- supervised registration methods on ScanNet and 3DMatch, even outperforming some older supervised methods. We also integrate our components into existing methods, showing their effectiveness.
