Helper to load the optimizer state from previous checkpoint Args: loaded_state: the state from the previous checkpoint

Helper to load the schedulers state from previous checkpoint Args: loaded_state: the state from the previous checkpoint

Return camera pose looking at a given center point. Analogous of gluLookAt function, using OpenCV camera convention.

https://github.com/kakaobrain/NeRF-Factory/blob/f61bb8744a5cb4820a4d968fb3bfbed777550f4a/src/model/mipnerf360/helper.py#L142 https://github.c

https://github.com/kakaobrain/NeRF-Factory/blob/f61bb8744a5cb4820a4d968fb3bfbed777550f4a/src/model/mipnerf360/helper.py#L136 https://github.c

Main function.

given data that follows the nerfstduio format such as the output from colmap or polycam, convert to a format that sdfstudio will ingest

Script to download existing datasets. We currently support the following datasets: - nerfstudio: Growing collection of real-world scenes. Use

Main function.

Main function.

Process video into a nerfstudio dataset.

Main function.

Export textured mesh

Main function.

Export mesh

Generate NuScenes dynamic object masks.

Create a tornado application for the websocket server.

mg_list: list of obj, sorted seq_len=5: -2,-1,ref,1,2

Compute mask iou between two binary masks. :param gt (numpy array, uint8): binary mask :param dt (numpy array, uint8): binary mask :r

If DDP, then return the .module. Otherwise, return the model.

normal consistency loss as monosdf Args: normal_pred (torch.Tensor): volume rendered normal normal_gt (torch.Tensor): monocular n

r""" This is the default pytorch collate function, but with support for nerfstudio types. All documentation below is copied straight over from

method: newCommandQueue

method: newDepthStencilStateWithDescriptor:

method: newSamplerStateWithDescriptor:

Returns the next eval image.

More info here: https://arxiv.org/pdf/2206.00665.pdf supplementary section A2 Depth Consistency Loss This function computes scale/shift requi

Modify camera intrinsics to follow a different convention. Coordinates of the center of the top-left pixels are by default: - (0.5, 0.5)

! @brief add to iterator @pre The iterator is initialized; i.e. `m_object != nullptr`. */

add to iterator

@brief add an object to an array @sa https://json.nlohmann.me/api/basic_json/operator+=/

! @brief subtract from iterator @pre The iterator is initialized; i.e. `m_object != nullptr`. */

Take an optimizer step using a grad scaler. Args: grad_scaler: GradScaler to use

Orientation loss proposed in Ref-NeRF. Loss that encourages that all visible normals are facing towards the camera.

Faster version of https://github.com/kakaobrain/NeRF-Factory/blob/f61bb8744a5cb4820a4d968fb3bfbed777550f4a/src/model/mipnerf360/helper.py#L117

tqdm but with parallel execution. Will essentially return res = [ function(arg) # default function(*arg) # if star_args is

Parse images from camera dict, which is useful when there are too many frames. TODO: images not in the camera dict can be used during validat

imgs: (B, 3, H, W) x: (B, L, patch_size**2 *3)

property: pixelFormat

Plot a set of images horizontally. Args: imgs: a list of NumPy or PyTorch images, RGB (H, W, 3) or mono (H, W). titles: a list of

compute convex hull of the given points

Set the necessary modules to get the network working.

extract 3D points/confidence from prediction head output