Neural Mesh Fusion: Unsupervised 3D Planar Surface Understanding
This addresses the problem of 3D scene understanding for computer vision applications, offering an unsupervised and efficient alternative to existing methods.
The paper tackles 3D planar surface understanding from multi-view images by introducing Neural Mesh Fusion (NMF), which jointly optimizes polygon mesh reconstruction and unsupervised planar segmentation without ground-truth supervision. Results show NMF achieves competitive performance with state-of-the-art methods while being significantly more computationally efficient than implicit neural rendering approaches.
This paper presents Neural Mesh Fusion (NMF), an efficient approach for joint optimization of polygon mesh from multi-view image observations and unsupervised 3D planar-surface parsing of the scene. In contrast to implicit neural representations, NMF directly learns to deform surface triangle mesh and generate an embedding for unsupervised 3D planar segmentation through gradient-based optimization directly on the surface mesh. The conducted experiments show that NMF obtains competitive results compared to state-of-the-art multi-view planar reconstruction, while not requiring any ground-truth 3D or planar supervision. Moreover, NMF is significantly more computationally efficient compared to implicit neural rendering-based scene reconstruction approaches.