← all datasets

Cross-View Completion

Emerging
1papers using it
2025first seen

Cross-View Completion is a benchmark that involves masked autoencoding of multiple views of the same scene to evaluate models on their ability to learn features for 3D understanding and perform tasks such as feedforward reconstruction, dense image matching, and relative pose estimation.

🔎 Find this dataset

Papers using Cross-View Completion (1)

Cross-View Completion — datasets — computer-vision