Checks whether a 1D tensor of indices is a permutation.

Check whether edge_index follows pytorch geometric graph edge format: a [2, N] torch.LongTensor.

Search k-NN for a 3D point cloud xyz and convert the output into torch_geometric's `edge_index`, `edge_attr` format. This search differs from

Using the top 7 prompt list to evaluate the mask of replica.

Read an HDF5 file and return its content as a Data object. NB: if relevant, a Batch object will be returned. :param f: h5 file path

Load a dictionary of torch.Tensor from an HDF5 file. :param f: h5 file path of h5py.File or h5py.Group or h5py.Dataset :param idx: int, list,

r"""Function that measures Learned Perceptual Image Patch Similarity (LPIPS). Arguments: x, y (torch.Tensor): the input tensors to co

Perform mask non-maximum suppression (NMS) on a set of masks based on their scores. Args: masks (torch.Tensor): has shape (num_m

postprocess segmentation results by merging small regions into neighbor regions with high affinity :param seg_labels: (s,), resulting labels

Fit an MLP to a point cloud. Assuming the point cloud mostly contains ground points, this function will train an MLP to model the ground surfa

Model the ground based on a trimmed point cloud carrying ground points only. At inference, a point is associated with its nearest neighbors in

code credit: SAM3D

Same as torch.randperm but leveraging numba on CPU. NB: slightly faster than `np.random.permutation(np.arange(n))`

Optionally recursive outlier search. The `xyz_query` and `xyz_search`. Search for points with less than `k_min` neighbors within a radius of `

Oversample partial neighborhoods with less than k points. Missing neighbors are indicated by the "-1" index. Remarks - Neighbors and di

Post-process a mesh to filter out floaters and disconnected parts

Prints content of DictConfig using Rich library and its tree structure. Args: cfg (DictConfig): Configuration composed by Hydra.

see: src/base/reconstruction.cc void Reconstruction::ReadImagesBinary(const std::string& path) void Reconstruction::WriteImagesBi

Taken from https://github.com/colmap/colmap/blob/dev/scripts/python/read_write_model.py

see: src/base/reconstruction.cc void Reconstruction::WriteCamerasBinary(const std::string& path) void Reconstruction::ReadCameras

Taken from https://github.com/colmap/colmap/blob/dev/scripts/python/read_write_model.py

see: src/base/reconstruction.cc void Reconstruction::ReadPoints3DBinary(const std::string& path) void Reconstruction::WritePoints

see: src/base/reconstruction.cc void Reconstruction::ReadPoints3DText(const std::string& path) void Reconstruction::WritePoints3D

Render the scene. Background tensor (bg_color) must be on GPU!

if string s represents an int.

Convert a 2D tensor of RGB colors int [0, 255] or float [0, 1] to HSV format. Credit: https://www.linuxtut.com/en/20819a90872275811439

Convert a tensor of RGB colors int[0, 255] or float [0, 1] to LAB colors. Reimplemented from: https://gist.github.com/manojpandey/f5ece71

Given an axis and a rotation angle, compute the rotation matrix using the Rodrigues formula. Source : https://en.wikipedia.org/wiki/Rodrigues

Save Data to HDF5 file. :param f: h5 file path of h5py.File or h5py.Group :param y_to_csr: bool Convert 'y' to CSR format

Save NAG to HDF5 file. :param path: :param y_to_csr: bool Convert 'y' to CSR format before saving. Only applies if

Save CSRData to HDF5 file. :param f: h5 file path of h5py.File or h5py.Group :param fp_dtype: torch dtype Data type to wh

Compress a 2D tensor with CSR format and save it in an already-open HDF5. :param x: 2D torch.Tensor :param f: h5 file path of h5py.File o

Save file in rank zero mode (only on one process in multi-GPU setup).

Save an image (probably a depthmap) to disk as a float32 TIFF.

Scatter implementation for mean normal orientation computation. When dealing with normals, we care more about the orientation than the sense.

Search for clusters and objects with 'void' semantic labels. IMPORTANT: By convention, we assume `y ∈ [0, num_classes-1]` ARE ALL

Returns a new CSRData which indexes `self` using entries in `idx`. Supports torch and numpy fancy indexing. :parameter idx: i

Model the ground as a single plane using RANSAC. Returns a callable taking an XYZ tensor as input and returning the pointwise elevation.

Lexicographic sort N-dimensional src points based on index first and the projection of the src values along a direction second.

Compute the subedges making up each edge between segments. These are needed for superedge features computation. This approach relies on heuris

Compute the target histogram for semantic segmentation. That is, for each cluster, the histogram of pointwise labels of its overlaps.

Reconstruct the list of `Data` objects that were passed to `Batch.from_data_list()`. This extends the behavior of PyG's `to_data_list

Transforms poses so principal components lie on XYZ axes. Args: poses: a (N, 3, 4) array containing the cameras' camera to world transforms.

3D data interactive visualization. :param input: `Data` or `NAG` object :param keys: `List(str)` or `str` By default, the following a