Co-Me: 視覚幾何Transformerのための信頼度誘導トークン統合
Co-Me: Confidence-Guided Token Merging for Visual Geometric Transformers
再学習なしで視覚幾何Transformerを高速化するため、軽量な信頼度予測器でトークンを不確実性順に並べ、低信頼トークンを統合する手法を提案。VGGTやPi3で最大21倍超の高速化を実現。
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: Yutian Chen, Yuheng Qiu, Ruogu Li, Ali Agha, Shayegan Omidshafiei, Jay Patrikar, Sebastian Scherer
分類: cs.CV, cs.RO
原文アブストラクト
We propose Confidence-Guided Token Merging (Co-Me), an acceleration mechanism for visual geometric transformers without retraining or finetuning the base model. Co-Me distilled a light-weight confidence predictor to rank tokens by uncertainty and selectively merge low-confidence ones, effectively reducing computation while maintaining spatial coverage. Compared to similarity-based merging or pruning, the confidence signal in Co-Me reliably indicates regions emphasized by the transformer, enabling substantial acceleration without degrading performance. Co-Me applies seamlessly to various multi-view and streaming visual geometric transformers, achieving speedups that scale with sequence length. When applied to VGGT and Pi3, Co-Me achieves up to 21.5x and 20.4x speedup, making visual geometric transformers practical for real-time 3D perception and reconstruction.
