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

↓ 1 callersFunctionapply_patch_shim

Crop images in the batch so that their dimensions are cleanly divisible by the specified patch size.

src/dataset/shims/patch_shim.py:30

↓ 1 callersFunctionavg_pool_nd

Create a 1D, 2D, or 3D average pooling module.

src/model/encoder/costvolume/ldm_unet/util.py:260

↓ 1 callersFunctionbatch_to_device

Move batch (dict) to device

src/model/LightGlue/lightglue/utils.py:55

↓ 1 callersMethodbuild

(self, config)

src/model/unidepth/unidepthv2/decoder.py:531

↓ 1 callersMethodbuild

(self, config)

src/model/unidepth/unidepthv2/unidepthv2.py:343

↓ 1 callersFunctionbuild_covariance

( scale: Float[Tensor, "*#batch 3"], rotation_xyzw: Float[Tensor, "*#batch 4"], )

src/model/encoder/common/gaussians.py:33

↓ 1 callersFunctioncamera_synchronization

Applies the proposed synchronization algorithm where the pairwise matrix is formed and iterative matrix multiplication is applied for synchronizat

src/flow_util.py:623

↓ 1 callersFunctioncenter_crop

( images: Float[Tensor, "*#batch c h w"], intrinsics: Float[Tensor, "*#batch 3 3"], shape: tuple[i

src/dataset/shims/crop_shim.py:25

↓ 1 callersMethodcheck_if_stop

evaluate stopping condition

src/model/LightGlue/lightglue/lightglue.py:640

↓ 1 callersFunctioncheckpoint_filter_fn

Remap FB checkpoints -> timm

src/model/unidepth/backbones/convnext.py:522

↓ 1 callersMethodclear_history

(self)

src/misc/benchmarker.py:39

↓ 1 callersFunctioncm_BlRdGn

Custom colormap: blue (-1) -> red (0.0) -> green (1).

src/model/LightGlue/lightglue/viz2d.py:22

↓ 1 callersFunctioncolor_function

( xy: Float[Tensor, "point 2"], )

src/visualization/drawing/lines.py:42

↓ 1 callersMethodcompile

( self, mode="reduce-overhead", static_lengths=[256, 512, 768, 1024, 1280, 1536] )

src/model/LightGlue/lightglue/lightglue.py:433

↓ 1 callersFunctioncompute_aabb

Compute an axis-aligned bounding box for the camera frustums.

src/visualization/drawing/cameras.py:123

↓ 1 callersMethodconfidence_threshold

scaled confidence threshold

src/model/LightGlue/lightglue/lightglue.py:626

↓ 1 callersFunctionconstruct_list_of_attributes

(num_rest: int)

src/model/ply_export.py:12

↓ 1 callersFunctionconvert_flow_to_mapping

(flow, output_channel_first=True)

src/flow_util.py:1070

↓ 1 callersMethodconvert_poses

( self, poses: Float[Tensor, "batch 18"], )

src/dataset/dataset_re10k.py:220

↓ 1 callersMethodconvert_poses

( self, poses: Float[Tensor, "batch 18"], )

src/dataset/dataset_dl3dv.py:223

↓ 1 callersMethodconvert_poses

( self, poses: Float[Tensor, "batch 18"], )

src/dataset/dataset_dl3dv_test.py:226

↓ 1 callersFunctionconvert_to_buffer

(module: nn.Module, persistent: bool = True)

src/misc/nn_module_tools.py:4

↓ 1 callersFunctioncoords_grid

(b, h, w, homogeneous=False, device=None)

src/model/encoder/costvolume/depth_predictor_multiview.py:10

↓ 1 callersMethodcopy_to

Copy current averaged parameters into given collection of parameters. Args: parameters: Iterable of `torch.nn.Parameter`

src/model/unidepth/utils/ema_torch.py:165

↓ 1 callersFunctioncreate_window

(window_size, channel)

src/loss/loss_multissim.py:48

↓ 1 callersFunctioncyan

(text: str)

src/main.py:33

↓ 1 callersMethoddecode_depth

(self, latents_16, rays, shapes)

src/model/unidepth/unidepthv2/decoder.py:259

↓ 1 callersFunctiondefault

(val, d)

src/model/unidepth/utils/misc.py:188

↓ 1 callersFunctiondepth_to_relative_disparity

Convert depth to relative disparity, where 0 is near and 1 is far

src/model/encoder/costvolume/conversions.py:17

↓ 1 callersFunctiondetect_msaa_pixels

( image: Float[Tensor, "batch 4 height width"], )

src/visualization/drawing/rendering.py:29

↓ 1 callersFunctiondownload_checkpoint

( run_id: str, download_dir: Path, version: str | None, )

src/misc/wandb_tools.py:11

↓ 1 callersFunctiondraw_cameras

( resolution: int, extrinsics: Float[Tensor, "batch 4 4"], intrinsics: Float[Tensor, "batch 3 3"],

src/visualization/drawing/cameras.py:14

↓ 1 callersFunctiondraw_label

Draw a black label on a white background with no border.

src/visualization/annotation.py:16

↓ 1 callersFunctiondrop_path

(x, drop_prob: float = 0.0, training: bool = False)

src/model/unidepth/backbones/metadinov2/drop_path.py:15

↓ 1 callersFunctiondrop_path

(x: torch.Tensor, drop_prob: float = 0.0, training: bool = False)

src/model/unidepth/layers/drop_path.py:5

↓ 1 callersMethodembed_rays

(self, rays, shapes)

src/model/unidepth/unidepthv2/decoder.py:112

↓ 1 callersMethodembed_rays

(self, rays, shapes)

src/model/unidepth/unidepthv2/decoder.py:239

↓ 1 callersFunctioneuler_to_matrix

( rotations: Float[Tensor, "*batch 3"], pattern: str, )

src/visualization/camera_trajectory/interpolation.py:96

↓ 1 callersFunctionevaluate

(cfg_dict: DictConfig)

src/scripts/compute_metrics.py:37

↓ 1 callersFunctionexport_ply

( extrinsics: Float[Tensor, "4 4"], means: Float[Tensor, "gaussian 3"], scales: Float[Tensor, "gau

src/model/ply_export.py:26

↓ 1 callersMethodextract_dense_map

(self, image)

src/model/LightGlue/lightglue/aliked.py:707

↓ 1 callersMethodfeature_add_position_list

(self, features_list, attn_splits, feature_channels)

src/model/encoder/encoder_costvolume.py:150

↓ 1 callersMethodfill_intrinsics

(self, x)

src/model/unidepth/unidepthv2/decoder.py:56

↓ 1 callersFunctionfilter_dog_point

(points, scales, angles, image_shape, nms_radius, scores=None)

src/model/LightGlue/lightglue/sift.py:17

↓ 1 callersFunctionfilter_matches

obtain matches from a log assignment matrix [Bx M+1 x N+1]

src/model/LightGlue/lightglue/lightglue.py:296

↓ 1 callersMethodforward

( self, probabilities: Float[Tensor, "*batch bucket"], num_samples: int, deter

src/model/encoder/common/sampler.py:11

↓ 1 callersMethodforward

(self, x: torch.Tensor)

src/model/unidepth/backbones/metadinov2/block.py:84

↓ 1 callersMethodforward

(self, x: Tensor)

src/model/unidepth/backbones/metadinov2/attention.py:48

↓ 1 callersMethodforward

Compute keypoints, scores, descriptors for image

src/model/LightGlue/lightglue/disk.py:28

↓ 1 callersMethodforward

( self, prediction: DecoderOutput, batch: BatchedExample, gaussians: Gaussians

src/loss/loss.py:29

↓ 1 callersMethodforward_nested

x_list contains a list of tensors to nest together and run

src/model/unidepth/backbones/metadinov2/block.py:229

↓ 1 callersFunctiongather_discrete_topk

( pdf: Float[Tensor, "*batch bucket"], num_samples: int, eps: float = torch.finfo(torch.float32).e

src/misc/discrete_probability_distribution.py:23

↓ 1 callersFunctiongaussian

(window_size, sigma)

src/loss/loss_multissim.py:44

↓ 1 callersFunctiongenerate_heterogeneous_index

Generate indices for all pairs except self-pairs.

src/misc/heterogeneous_pairings.py:9

↓ 1 callersFunctiongenerate_rotation_coordinate_frame

Generate a coordinate frame where the Y direction is normal to the plane defined by unit vectors a and b. The other axes are arbitrary.

src/visualization/camera_trajectory/interpolation.py:64

↓ 1 callersFunctiongenerate_shift_window_attn_mask

(input_resolution, window_size_h, window_size_w, shift_size_h, shift_size_

src/model/encoder/multiview_transformer.py:69

↓ 1 callersFunctiongenerate_spin

( num_frames: int, device: torch.device, elevation: float, radius: float, )

src/visualization/camera_trajectory/spin.py:9

↓ 1 callersFunctiongenerate_wobble

( extrinsics: Float[Tensor, "*#batch 4 4"], radius: Float[Tensor, "*#batch"], t: Float[Tensor, " t

src/visualization/camera_trajectory/wobble.py:26

↓ 1 callersFunctiongeometric_progression

(a, r, n)

src/model/unidepth/utils/misc.py:264

↓ 1 callersFunctionget_2d_sincos_pos_embed

This function initializes a grid and generates a 2D positional embedding using sine and cosine functions. It is a wrapper of get_2d_sincos_po

src/flow_util.py:281

↓ 1 callersFunctionget_2d_sincos_pos_embed_from_grid

This function generates a 2D positional embedding from a given grid using sine and cosine functions. Args: - embed_dim: The embedding di

src/flow_util.py:255

↓ 1 callersMethodget_adapted_features

(self, features_flat, splits)

src/model/unidepth/unidepthv2/decoder.py:373

↓ 1 callersFunctionget_branges_scales

(x, sample_drop_ratio=0.0)

src/model/unidepth/backbones/metadinov2/block.py:138

↓ 1 callersMethodget_current_decay

(self)

src/model/unidepth/utils/ema_torch.py:132

↓ 1 callersFunctionget_data_shim

Get functions that modify the batch. It's sometimes necessary to modify batches outside the data loader because GPU computations are required to m

src/dataset/data_module.py:17

↓ 1 callersMethodget_data_shim

The default shim doesn't modify the batch.

src/model/encoder/encoder.py:27

↓ 1 callersFunctionget_encoder

(cfg: EncoderCfg)

src/model/encoder/__init__.py:15

↓ 1 callersFunctionget_lpips

(device: torch.device)

src/evaluation/metrics.py:23

↓ 1 callersFunctionget_num_layer_for_convnext

Divide [3, 3, 27, 3] layers into 12 groups; each group is three consecutive blocks, including possible neighboring downsample layers; ada

src/model/unidepth/backbones/convnext.py:24

↓ 1 callersFunctionget_num_layer_for_convnext

Divide [3, 3, 27, 3] layers into 12 groups; each group is three consecutive blocks, including possible neighboring downsample layers; ada

src/model/unidepth/backbones/convnext2.py:26

↓ 1 callersFunctionget_num_layer_for_swin

(var_name, num_max_layer, layers_per_stage)

src/model/unidepth/utils/misc.py:117

↓ 1 callersMethodget_output_dim

Same as above. The default for input_dims is 3 for 3D applications which use harmonic embedding for positional encoding, so t

src/flow_util.py:601

↓ 1 callersMethodget_output_dim_static

Utility to help predict the shape of the output of `forward`. Args: input_dims: length of the last dimension of the inpu

src/flow_util.py:585

↓ 1 callersFunctionget_parameter_groups

(model, lr, wd=1e-5, ld=0.9, skip_list=())

src/model/unidepth/backbones/dinov2.py:37

↓ 1 callersFunctionget_parameter_groups

(model, lr, wd=1e-5, ld=0.9, skip_list=None)

src/model/unidepth/backbones/convnext.py:56

↓ 1 callersFunctionget_parameter_groups

(model, lr, wd=1e-5, ld=0.9, skip_list=())

src/model/unidepth/backbones/convnext2.py:57

↓ 1 callersFunctionget_rank

()

src/model/unidepth/utils/distributed.py:27

↓ 1 callersMethodget_scale_multiplier

( self, intrinsics: Float[Tensor, "*#batch 3 3"], pixel_size: Float[Tensor, "*#batch 2

src/model/encoder/common/gaussian_adapter.py:100

↓ 1 callersMethodget_scaleshift

(self, x, h, w)

src/model/encoder/encoder_costvolume.py:167

↓ 1 callersMethodget_scaleshift

(self, x)

src/model/unidepth/unidepthv2/decoder.py:130

↓ 1 callersMethodget_step

(self)

src/misc/step_tracker.py:21

↓ 1 callersFunctionget_view_sampler

( cfg: ViewSamplerCfg, stage: Stage, overfit: bool, cameras_are_circular: bool, step_track

src/dataset/view_sampler/__init__.py:26

↓ 1 callersFunctionhpoints_to_points

(hpoints)

src/flow_util.py:337

↓ 1 callersFunctionimage_grid

(imgs, rows, cols)

src/model/unidepth/utils/visualization.py:39

↓ 1 callersMethodindex

(self)

src/dataset/dataset_re10k.py:270

↓ 1 callersMethodinfer

(self, rgbs: torch.Tensor, intrinsics=None)

src/model/unidepth/unidepthv2/unidepthv2.py:225

↓ 1 callersMethodinit_latents

(self, features, shapes)

src/model/unidepth/unidepthv2/decoder.py:310

↓ 1 callersMethodinit_weights

(self)

src/model/unidepth/backbones/dinov2.py:256

↓ 1 callersFunctioninterpolate_circular

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

src/visualization/camera_trajectory/interpolation.py:162

↓ 1 callersFunctioninterpolate_pivot_parameters

( initial: Float[Tensor, "*#batch 5"], final: Float[Tensor, "*#batch 5"], t: Float[Tensor, " time_

src/visualization/camera_trajectory/interpolation.py:190

↓ 1 callersMethodinterpolate_pos_encoding

(self, x, w, h)

src/model/unidepth/backbones/dinov2.py:263

↓ 1 callersFunctionintersect_rays

Compute the least-squares intersection of rays. Uses the math from here: https://math.stackexchange.com/a/1762491/286022

src/visualization/camera_trajectory/interpolation.py:19

↓ 1 callersFunctionintersect_rays

Compute the least-squares intersection of rays. Uses the math from here: https://math.stackexchange.com/a/1762491/286022

src/geometry/projection.py:176

↓ 1 callersFunctionis_main_process

()

src/model/unidepth/utils/distributed.py:39

↓ 1 callersFunctionlift_to_3d

Calculate the 3D positions that correspond to the specified 2D points on the epipolar lines defined by the origins and directions. The extrinsics

src/geometry/epipolar_lines.py:264

↓ 1 callersFunctionload_checkpoint_swin

(model, checkpoint_model)

src/model/unidepth/utils/misc.py:246

↓ 1 callersFunctionload_image

( path: Union[Path, str], )

src/misc/image_io.py:71

↓ 1 callersMethodmap_pdf_to_opacity

( self, pdf: Float[Tensor, " *batch"], global_step: int, )

src/model/encoder/encoder_costvolume.py:174

↓ 1 callersFunctionmap_tensor

(input_, func: Callable)

src/model/LightGlue/lightglue/utils.py:41

↓ 1 callersMethodmasked_forward

(self, desc0, desc1, encoding0, encoding1, mask0, mask1)

src/model/LightGlue/lightglue/lightglue.py:250

← previous next →201–300 of 843, ranked by callers