Get 3D object shape from sparse views without camera poses by aggregating 2D image feature with LEAP LEAP: Liberate Sparse-view 3D Modeling from Camera Poses arXiv paper abstract https://arxiv.org/abs/2310.01410 arXiv PDF paper https://arxiv.org/pdf/2310.01410.pdf Project page https://hwjiang1510.github.io/LEAP … for multi-view 3D modeling … Existing approaches predominantly assume access to accurate camera…

Segment object with only image labels and paying attention to entire parts of objects with COMNet COMNet: Co-Occurrent Matching for Weakly Supervised Semantic Segmentation arXiv paper abstract https://arxiv.org/abs/2309.16959 arXiv PDF paper https://arxiv.org/pdf/2309.16959.pdf Image-level weakly supervised semantic segmentation is a challenging task that has been deeply studied in recent years. Most…

Segment scene in new domain using multiple sets of discrepancy features with Feng Open Compound Domain Adaptation with Object Style Compensation for Semantic Segmentation arXiv paper abstract https://arxiv.org/abs/2309.16127 arXiv PDF paper https://arxiv.org/pdf/2309.16127.pdf Many methods of semantic image segmentation have borrowed the success of open compound domain adaptation. … paper proposes…

Get 3D scene with fewer images by adapting scene priors trained on large datasets with NFP 3D Reconstruction with Generalizable Neural Fields using Scene Priors arXiv paper abstract https://arxiv.org/abs/2309.15164 arXiv PDF paper https://arxiv.org/pdf/2309.15164.pdf Project page https://oasisyang.github.io/neural-prior High-fidelity 3D scene reconstruction has been substantially advanced by recent progress in neural fields.

Real-time unsupervised video object segmentation by training with images and optical flow with TMO Treating Motion as Option with Output Selection for Unsupervised Video Object Segmentation arXiv paper abstract https://arxiv.org/abs/2309.14786 arXiv PDF paper https://arxiv.org/pdf/2309.14786.pdf GitHub https://github.com/suhwan-cho/TMO Unsupervised video object segmentation (VOS) is a task that aims to detect the most…

Segment scene in new domain with one image using adversarial refinement with SITTA-SEG Single Image Test-Time Adaptation for Segmentation arXiv paper abstract https://arxiv.org/abs/2309.14052 arXiv PDF paper https://arxiv.org/pdf/2309.14052.pdf Project page https://klarajanouskova.github.io/sitta-seg Test-Time Adaptation (TTA) methods improve the robustness of deep neural networks to domain shift on a variety of tasks such…

Get 3D object from scene by lifting Segment Anything Model masks into 3D field with NOC NOC: High-Quality Neural Object Cloning with 3D Lifting of Segment Anything arXiv paper abstract https://arxiv.org/abs/2309.12790 arXiv PDF paper https://arxiv.org/pdf/2309.12790.pdf With the development of the neural field, reconstructing the 3D model of a target object…

Real-time video object segmentation by using global assignment and semantic features with TCOVIS TCOVIS: Temporally Consistent Online Video Instance Segmentation arXiv paper https://arxiv.org/abs/2309.11857 arXiv PDF paper https://arxiv.org/pdf/2309.11857.pdf … progress has been made in video instance segmentation (VIS), with many offline and online methods achieving state-of-the-art performance … online methods are…

3D point cloud segmentation with few examples using shared geometric components with GFS-3DSeg_GWs Generalized Few-Shot Point Cloud Segmentation Via Geometric Words arXiv paper abstract https://arxiv.org/abs/2309.11222 arXiv PDF paper https://arxiv.org/pdf/2309.11222.pdf Existing fully-supervised point cloud segmentation methods suffer in the dynamic testing environment with emerging new classes. Few-shot point cloud segmentation algorithms…