Test3R: Learning to Reconstruct 3D at Test Time
This work addresses geometric inconsistency in 3D reconstruction for computer vision applications, offering a universal and cost-effective solution that is incremental in its approach.
The paper tackles the problem of limited geometric consistency in 3D reconstruction by introducing Test3R, a test-time learning technique that optimizes networks via a self-supervised objective to maximize consistency between reconstructions from image triplets, resulting in significant performance improvements over previous state-of-the-art methods.
Dense matching methods like DUSt3R regress pairwise pointmaps for 3D reconstruction. However, the reliance on pairwise prediction and the limited generalization capability inherently restrict the global geometric consistency. In this work, we introduce Test3R, a surprisingly simple test-time learning technique that significantly boosts geometric accuracy. Using image triplets ($I_1,I_2,I_3$), Test3R generates reconstructions from pairs ($I_1,I_2$) and ($I_1,I_3$). The core idea is to optimize the network at test time via a self-supervised objective: maximizing the geometric consistency between these two reconstructions relative to the common image $I_1$. This ensures the model produces cross-pair consistent outputs, regardless of the inputs. Extensive experiments demonstrate that our technique significantly outperforms previous state-of-the-art methods on the 3D reconstruction and multi-view depth estimation tasks. Moreover, it is universally applicable and nearly cost-free, making it easily applied to other models and implemented with minimal test-time training overhead and parameter footprint. Code is available at https://github.com/nopQAQ/Test3R.