CameraModel(model_id=0, model_name="SIMPLE_PINHOLE", num_params=3), CameraModel(model_id=1, model_name="PINHOLE", num_params=4), CameraMo

# original implementation from https://arxiv.org/abs/2307.15250 Criterion for point only

Store and Merge the 3D lines output from the model (by remove line with high uncertainty) lines3D: (N, 6) points3D: (N, 3) Visualize

Multi-head attention to increase model expressivitiy

Wrapper around torch.optim + learning rate

Original from: https://github.com/nianticlabs/ace Compute per-pixel reprojection loss using different configurable approaches. - tanh:

SuperPoint Convolutional Detector and Descriptor SuperPoint: Self-Supervised Interest Point Detection and Description. Daniel DeTone, Tomasz