1

VeloDepth is an online video depth method that propagates depth features and predictions across frames using flow-based warping with learned residual corrections. This design enforces temporal consistency while achieving stable depth at significantly higher speed with strong zero-shot accuracy.

UniK3D is capable of estimating metric 3D scenes across domains and for any camera from solely single images. UniK3D directly predicts metric 3D points from the input image at inference time without any additional information.

UniDepth is capable of reconstructing metric 3D scenes from solely single images across domains. UniDepth directly predicts metric 3D points from the input image at inference time without any additional information.