Convert depth map to color visualization using OpenCV colormap. Args: depth_map (np.ndarray): Depth map (H, W) vmin (float):

Convert depth map to point cloud (pure Tensor version, no point filtering) Args: depth: torch.Tensor, shape (H, W) or (B, H, W), dep

this will perform the index select, cat the tensors, and provide the attn_bias from cache

Compute depth validity for each patch Args: depth_values: [B, 1, H_img, W_img] raw depth map H_patch, W_patch: patch grid di

Perform patch masking based on depth validity Args: x: [B, N, D] input features (after patch embedding) patch_num_h: int, he

Convert depth map to point cloud (pure Tensor version, no point filtering) Args: depth: torch.Tensor, shape (H, W) or (B, H, W), dep

Flattens a nested dictionary into a single-level dictionary, with keys as tuples.

x_list contains a list of tensors to nest together and run

Calculate lr decay rate for different ViT blocks. Args: name (string): parameter name. lr_decay_rate (float): base lr decay r

Load depth map from PNG file (16-bit) and convert to meters. Args: depth_path (str): Path to depth image scale (float): Scal

Load camera intrinsics and normalize by image dimensions. Args: intrinsics_path (str): Path to intrinsics file (.txt or .json)

Main function with argument parsing and execution.

UV with left-top corner as (-width / diagonal, -height / diagonal) and right-bottom corner as (width / diagonal, height / diagonal)

Load and preprocess RGB image. Args: image_path (str): Path to RGB image device (torch.device): Device to load tensor on

Args: img_size (int, tuple): input image size patch_size (int, tuple): patch size in_chans (int): number