Functions — github.com/cvlab-kaist/PF3plat

hub / github.com/cvlab-kaist/PF3plat / functions

Functions843 in github.com/cvlab-kaist/PF3plat

⨍ Functions 843◇ Types & classes 197↳ Endpoints 2

r"""Move internal buffers of the ExponentialMovingAverage to `device`. Args: device: like `device` argument to `torch.Tensor.to`

src/model/unidepth/utils/ema_torch.py:249

↓ 47 callersMethodclone

(self)

src/model/types.py:13

↓ 38 callersFunctionadd_label

( image: Float[Tensor, "3 width height"], label: str, font: Path = Path("assets/Inter-Regular.otf"

src/visualization/annotation.py:38

↓ 25 callersFunctionhcat

Shorthand for a horizontal linear concatenation.

src/visualization/layout.py:149

↓ 22 callersFunctionadd_border

( image: Float[Tensor, "channel height width"], border: int = 8, color: Color = 1, )

src/visualization/layout.py:193

↓ 21 callersFunctionconv_nd

Create a 1D, 2D, or 3D convolution module.

src/model/encoder/costvolume/ldm_unet/util.py:240

↓ 20 callersFunctionvcat

Shorthand for a horizontal linear concatenation.

src/visualization/layout.py:171

↓ 17 callersMethoddevice

(self)

src/model/unidepth/utils/sht.py:1436

↓ 14 callersMethodgather

Gather from the target according to the specified index. Handle the broadcasting needed for the gather to work. See the comments for the actua

src/model/encoder/common/sampler.py:26

↓ 13 callersFunctiondraw_points

( image: Float[Tensor, "3 height width"], points: Vector, color: Vector = [1, 1, 1], radius: S

src/visualization/drawing/points.py:13

↓ 13 callersMethodload_state_dict

r"""Loads the ExponentialMovingAverage state. Args: state_dict (dict): EMA state. Should be an object returned fr

src/model/unidepth/utils/ema_torch.py:287

↓ 12 callersFunctionnormalization

Make a standard normalization layer. :param channels: number of input channels. :return: an nn.Module for normalization.

src/model/encoder/costvolume/ldm_unet/util.py:199

↓ 11 callersFunctionprep_image

(image: FloatImage)

src/misc/image_io.py:38

↓ 11 callersFunctionsplit_feature

(feature, num_splits=2, channel_last=False, )

src/model/encoder/multiview_transformer.py:6

↓ 10 callersMethod__init__

(self, channels, use_conv, dims=2, out_channels=None, padding=1, downsample_3ddim=False, # downsample all

src/model/encoder/costvolume/ldm_unet/unet.py:102

↓ 10 callersFunctiondraw_lines

( image: Float[Tensor, "3 height width"], start: Vector, end: Vector, color: Vector, width

src/visualization/drawing/lines.py:13

↓ 9 callersMethodlog_image

( self, key: str, images: list[Any], step: Optional[int] = None, **kwa

src/misc/LocalLogger.py:35

↓ 9 callersFunctionzero_module

Zero out the parameters of a module and return it.

src/model/encoder/costvolume/ldm_unet/util.py:174

↓ 8 callersFunctioncompute_psnr

( ground_truth: Float[Tensor, "batch channel height width"], predicted: Float[Tensor, "batch channel h

src/evaluation/metrics.py:12

↓ 8 callersMethodextract

Perform extraction with online resizing

src/model/LightGlue/lightglue/utils.py:137

↓ 7 callersMethod__init__

(self, allow_flash: bool)

src/model/LightGlue/lightglue/lightglue.py:91

↓ 7 callersMethoddump

(self, path: Path)

src/misc/benchmarker.py:25

↓ 7 callersFunctionget_distinct_color

(index: int)

src/visualization/colors.py:30

↓ 7 callersFunctionget_world_rays

( coordinates: Float[Tensor, "*#batch dim"], extrinsics: Float[Tensor, "*#batch dim+2 dim+2"], int

src/geometry/projection.py:91

↓ 7 callersFunctionsample_image_grid

Get normalized (range 0 to 1) coordinates and integer indices for an image.

src/geometry/projection.py:117

↓ 6 callersMethod__init__

(self, dim_in, dim_out)

src/model/encoder/costvolume/ldm_unet/attention.py:36

↓ 6 callersFunctionbatch_project_to_other_img

Project pixels of one image to the other. Args: kpi: BxNx2 coordinates in pixels of image i di: BxN, corresponding depths of

src/flow_util.py:195

↓ 6 callersFunctiongenerate_rays

( camera_intrinsics, image_shape, noisy = False )

src/model/unidepth/utils/geometric.py:13

↓ 6 callersFunctionget_cfg

()

src/global_cfg.py:8

↓ 6 callersFunctionnn_gather

(points, indices)

src/flow_util.py:907

↓ 6 callersFunctionproject

(points: Float[Tensor, "*batch 3"])

src/visualization/drawing/cameras.py:52

↓ 6 callersFunctionsave_image

Save an image. Assumed to be in range 0-1.

src/misc/image_io.py:57

↓ 6 callersMethodself_attn

( self, x: torch.Tensor, attn_bias: torch.Tensor | None = None, pos_embed: tor

src/model/unidepth/layers/attention.py:249

↓ 5 callersMethod__init__

Args: radius: soft detection radius, kernel size is (2 * radius + 1) top_k: top_k > 0: return top k keypoints

src/model/LightGlue/lightglue/aliked.py:95

↓ 5 callersMethod_get_parameters

( self, parameters: Optional[Iterable[torch.nn.Parameter]] )

src/model/unidepth/utils/ema_torch.py:111

↓ 5 callersFunctionapply_color_map_to_image

( image: Float[Tensor, "*batch height width"], color_map: str = "inferno", )

src/visualization/color_map.py:22

↓ 5 callersFunctionbox

( image: Float[Tensor, "3 height width"], )

src/model/encoder/visualization/encoder_visualizer_costvolume.py:27

↓ 5 callersFunctioncompute_geodesic_distance_from_two_matrices

(m1, m2)

src/flow_util.py:20

↓ 5 callersFunctioncompute_lpips

( ground_truth: Float[Tensor, "batch channel height width"], predicted: Float[Tensor, "batch channel h

src/evaluation/metrics.py:28

↓ 5 callersFunctioncompute_ssim

( ground_truth: Float[Tensor, "batch channel height width"], predicted: Float[Tensor, "batch channel h

src/evaluation/metrics.py:37

↓ 5 callersMethodforward

( self, context: BatchedViews, deterministic: bool, )

src/model/encoder/encoder.py:20

↓ 5 callersMethodno_weight_decay

(self)

src/model/unidepth/backbones/convnext.py:513

↓ 5 callersFunctionsanitize_scalar

(scalar: Scalar, device: torch.device)

src/visualization/drawing/types.py:44

↓ 5 callersFunctionsanitize_vector

( vector: Vector, dim: int, device: torch.device, )

src/visualization/drawing/types.py:18

↓ 5 callersFunctionunproject_frustum_corners

( extrinsics: Float[Tensor, "batch 4 4"], intrinsics: Float[Tensor, "batch 3 3"], depth: Float[Ten

src/visualization/drawing/cameras.py:168

↓ 4 callersMethod__init__

( self, config, )

src/model/unidepth/unidepthv2/decoder.py:350

↓ 4 callersFunction_compute_point_projection

( xyz: Float[Tensor, "*#batch 3"], t: Float[Tensor, "*#batch"], intrinsics: Float[Tensor, "*#batch

src/geometry/epipolar_lines.py:134

↓ 4 callersFunction_intersect_image_coordinate

Compute the intersection of the projection of a camera-space ray with a line that's parallel to the image frame, either horizontally or vertically

src/geometry/epipolar_lines.py:55

↓ 4 callersMethodattn

( self, x: torch.Tensor, attn_bias: torch.Tensor | None = None, context: torch

src/model/unidepth/layers/attention.py:108

↓ 4 callersFunctioncolorize

( value, vmin, vmax = None, cmap = "magma_r" )

src/model/unidepth/utils/visualization.py:17

↓ 4 callersFunctionflat_interpolate

( flat_tensor, old, new, antialias = True, mode= "bilinear", )

src/model/unidepth/utils/geometric.py:225

↓ 4 callersMethodget_bound

( self, bound: Literal["near", "far"], num_views: int, )

src/dataset/dataset_re10k.py:253

↓ 4 callersMethodget_bound

( self, bound: Literal["near", "far"], num_views: int, )

src/dataset/dataset_dl3dv.py:256

↓ 4 callersMethodget_bound

( self, bound: Literal["near", "far"], num_views: int, )

src/dataset/dataset_dl3dv_test.py:259

↓ 4 callersFunctionget_conv

( inplanes, planes, kernel_size=3, stride=1, padding=1, bias=False, conv_type="con

src/model/LightGlue/lightglue/aliked.py:352

↓ 4 callersFunctionget_fov

(intrinsics: Float[Tensor, "batch 3 3"])

src/geometry/projection.py:233

↓ 4 callersFunctionhomogenize_points

Convert batched points (xyz) to (xyz1).

src/geometry/projection.py:9

↓ 4 callersFunctionkabsch_algorithm

Runs the weighted kabsh algorithm ... Args: corr_P (FloatTensor): pointcloud P (batch, N, 3) corr_Q (FloatTensor): pointcloud Q (

src/flow_util.py:888

↓ 4 callersFunctionload_image

(path: Path, resize: int = None, **kwargs)

src/model/LightGlue/lightglue/utils.py:124

↓ 4 callersFunctionload_typed_config

( cfg: DictConfig, data_class: Type[T], extra_type_hooks: dict = {}, )

src/config.py:61

↓ 4 callersMethodmap_

(self, func: Callable, x0: torch.Tensor, x1: torch.Tensor)

src/model/LightGlue/lightglue/lightglue.py:192

↓ 4 callersFunctionmerge_splits

(splits, num_splits=2, channel_last=False, )

src/model/encoder/multiview_transformer.py:35

↓ 4 callersFunctionnormalize

(a: Float[Tensor, "*#batch dim"])

src/visualization/camera_trajectory/interpolation.py:51

↓ 4 callersFunctionpad_to_length

(x: torch.Tensor, length: int)

src/model/LightGlue/lightglue/lightglue.py:39

↓ 4 callersFunctionprocess_vector

(vector)

src/geometry/projection.py:236

↓ 4 callersFunctionset_cfg

(new_cfg: DictConfig)

src/global_cfg.py:13

↓ 4 callersFunctionunproject

Unproject 2D camera coordinates with the given Z values.

src/geometry/projection.py:74

↓ 3 callersMethod__init__

(self, use_dropout=False, attention_dropout=0.1)

src/model/encoder/aggregation.py:54

↓ 3 callersMethod__init__

Args: in_chans: Number of input image channels. num_classes: Number of classes for classification head. g

src/model/unidepth/backbones/convnext.py:312

↓ 3 callersMethod__init__

(self, dim)

src/model/unidepth/backbones/convnext2.py:145

↓ 3 callersMethod__init__

( self, num_pos_feats=64, temperature=10000, normalize=False, scale=None )

src/model/unidepth/utils/positional_embedding.py:16

↓ 3 callersMethod__init__

( self, num_pos_feats=64, temperature=10000, normalize=False, scale=None )

src/model/unidepth/layers/positional_encoding.py:17

↓ 3 callersFunction_make_dinov2_model

( *, arch_name: str = "vit_large", img_size: int = 518, patch_size: int = 14, init_values:

src/model/unidepth/backbones/dinov2.py:407

↓ 3 callersFunction_many

(fn)

src/model/unidepth/utils/misc.py:218

↓ 3 callersFunctionapply_augmentation_shim

Randomly augment the training images.

src/dataset/shims/augmentation_shim.py:24

↓ 3 callersFunctionapply_crop_shim

Crop images in the example.

src/dataset/shims/crop_shim.py:87

↓ 3 callersFunctioncheckpoint

Evaluate a function without caching intermediate activations, allowing for reduced memory at the expense of extra compute in the backward pas

src/model/encoder/costvolume/ldm_unet/util.py:102

↓ 3 callersMethodcross_attn

( self, x: torch.Tensor, attn_bias: torch.Tensor | None = None, context: torch

src/model/unidepth/layers/attention.py:210

↓ 3 callersMethoddata_shim

(batch: BatchedExample)

src/model/encoder/encoder_costvolume.py:580

↓ 3 callersFunctiondefault

(val, d)

src/model/encoder/costvolume/ldm_unet/attention.py:17

↓ 3 callersFunctiondrawpoint

(img1, pts1, img2, pts2, colors)

src/flow_util.py:1173

↓ 3 callersMethodforward

(self, x)

src/model/encoder/costvolume/ldm_unet/util.py:223

↓ 3 callersFunctiongenerate_coordinate_frame

Generate a coordinate frame given perpendicular, unit-length Y and Z vectors.

src/visualization/camera_trajectory/interpolation.py:55

↓ 3 callersFunctionget_dataset

( cfg: DatasetCfg, stage: Stage, step_tracker: StepTracker | None, )

src/dataset/__init__.py:22

↓ 3 callersMethodget_generator

(self, loader_cfg: DataLoaderStageCfg)

src/dataset/data_module.py:83

↓ 3 callersMethodget_persistent

(self, loader_cfg: DataLoaderStageCfg)

src/dataset/data_module.py:80

↓ 3 callersMethodget_resblock

(self, c_in, c_out, conv_type, mask)

src/model/LightGlue/lightglue/aliked.py:695

↓ 3 callersFunctionmax_pool

(x)

src/model/LightGlue/lightglue/superpoint.py:56

↓ 3 callersFunctionproject_rays

( origins: Float[Tensor, "*#batch 3"], directions: Float[Tensor, "*#batch 3"], extrinsics: Float[T

src/geometry/epipolar_lines.py:157

↓ 3 callersFunctionrender_cuda

( extrinsics: Float[Tensor, "batch 4 4"], intrinsics: Float[Tensor, "batch 3 3"], near: Float[Tens

src/model/decoder/cuda_splatting.py:47

↓ 3 callersMethodrender_video_generic

( self, batch: BatchedExample, trajectory_fn: TrajectoryFn, name: str,

src/model/model_wrapper.py:699

↓ 3 callersFunctionrotate_sh

( sh_coefficients: Float[Tensor, "*#batch n"], rotations: Float[Tensor, "*#batch 3 3"], )

src/misc/sh_rotation.py:10

↓ 3 callersMethodsample

( self, scene: str, extrinsics: Float[Tensor, "view 4 4"], intrinsics: Float[T

src/dataset/view_sampler/view_sampler.py:36

↓ 3 callersMethodset_shapes

(self, shapes: tuple[int, int])

src/model/unidepth/unidepthv2/decoder.py:236

↓ 3 callersMethodupdate

Update currently maintained parameters. Call this every time the parameters are updated, such as the result of the `optimize

src/model/unidepth/utils/ema_torch.py:139

↓ 2 callersFunctionNormalize

(in_channels)

src/model/encoder/costvolume/ldm_unet/attention.py:74

↓ 2 callersMethod__init__

(self, target_dim, n_harmonic_functions=10, append_input=True)

src/flow_util.py:611

↓ 2 callersMethod__init__

(self, d_model=256, nhead=1, attention_type='swin',

src/model/encoder/multiview_transformer.py:309

↓ 2 callersMethod__init__

( self, hidden_dim, num_layers: int = 2, expansion: int = 4, layer_sca

src/model/unidepth/layers/upsample.py:14

↓ 2 callersMethod__init__

( self, dim: int, num_heads: int = 4, expansion: int = 4, dropout: flo

src/model/unidepth/layers/attention.py:82

next →1–100 of 843, ranked by callers