Functions — github.com/XunshanMan/MVGFormer

hub / github.com/XunshanMan/MVGFormer / functions

Functions561 in github.com/XunshanMan/MVGFormer

⨍ Functions 561◇ Types & classes 78

↓ 2 callersFunctionoks_iou

Calculate oks ious. Args: g: Ground truth keypoints. d: Detected keypoints. a_g: Area of the ground truth object.

lib/core/nms.py:51

↓ 2 callersFunctionposes_to_bone_length

(poses_3d, n_person, conv_J2B)

process/extract_dataset.py:169

↓ 2 callersFunctionproject_pose

(x, camera)

lib/utils/cameras.py:220

↓ 2 callersFunctionproject_pose_batch

(x, camera, deal_distortion=True)

lib/utils/cameras.py:210

↓ 2 callersMethodproject_ref_points

(self, reference_points, meta, nviews, batch_size, nbins, device)

lib/models/dq_decoder.py:331

↓ 2 callersFunctionreplace_gt_bboxes_with_cnn

Replace ground truth bounding boxes with boxes from a CNN detector.

lib/mvn/datasets/human36m_preprocessing/generate-labels-npy-multiview.py:110

↓ 2 callersFunctionsave_batch_heatmaps_multi

batch_image: [batch_size, channel, height, width] batch_heatmaps: ['batch_size, num_joints, height, width] file_name: saved file name

lib/utils/vis.py:392

↓ 2 callersFunctionsave_batch_image_with_joints_multi

batch_image: [batch_size, channel, height, width] batch_joints: [batch_size, num_person, num_joints, 3], batch_joints_vis: [batch_size, n

lib/utils/vis.py:349

↓ 2 callersFunctionsave_checkpoint

(states, is_best, output_dir, filename='checkpoint.pth.tar')

lib/utils/utils.py:144

↓ 2 callersFunctionsave_image_from_loader

(loader, save_dir, prefix)

process/extract_dataset.py:149

↓ 2 callersFunctionsquare_the_bbox

(bbox)

lib/mvn/datasets/human36m_preprocessing/generate-labels-npy-multiview.py:87

↓ 2 callersFunctionstr_to_number

(x)

lib/utils/string_parser.py:6

↓ 2 callersFunctionunfold_camera_param

(camera, device=None)

lib/utils/cameras.py:97

↓ 2 callersMethodupdate

(self, value, n=1)

lib/utils/misc.py:63

↓ 2 callersMethodupdate

(self, value, n=1)

lib/models/util/misc.py:99

↓ 2 callersMethodupdate_after_resize

(self, image_shape, new_image_shape)

lib/mvn/utils/multiview.py:33

↓ 2 callersFunctionvalidate_3d

(config, model, loader, output_dir, threshold, num_views=5, device="cpu", epoch=0, frame_id=None)

lib/core/function.py:329

↓ 1 callersFunctionPose3D_inference_torch

ONE PERSON EACH TIME [Torch Version with Auto Differentiate] The main procedure of structural triangulation & step contraint algorit

lib/structural/structural_triangulation.py:330

↓ 1 callersFunctionPose3D_inference_torch_batch

Batch version. Faster. [Torch Version with Auto Differentiate] The main procedure of structural triangulation & step contraint algor

lib/structural/structural_triangulation.py:209

↓ 1 callersFunctionST_SCA

Structural Triangulation with step contrain algorithm. When n_step == 1, this is pure ST without SCA.

lib/structural/structural_triangulation.py:509

↓ 1 callersMethod__getitem__

(self, idx)

lib/dataset/h36m.py:217

↓ 1 callersMethod__init__

(self, *args, joint_regressor_extra='data/J_regressor_extra.npy', **kwargs)

lib/smpl/smpl.py:78

↓ 1 callersMethod__init__

(self, cfg, decoder_layer, num_layers, return_intermediate=False)

lib/models/mvp_decoder.py:268

↓ 1 callersMethod__init__

(self)

lib/models/dq_transformer.py:69

↓ 1 callersMethod_check

(self)

lib/smpl/smpl_common.py:50

↓ 1 callersFunction_check_size_scale_factor

(dim, size, scale_factor)

lib/models/util/misc.py:50

↓ 1 callersMethod_create_sub_modules

create theta discriminator for 23 joint

lib/smpl/smpl_common.py:199

↓ 1 callersMethod_eval_list_to_ap

(eval_list, total_gt, threshold)

lib/dataset/h36m.py:286

↓ 1 callersMethod_eval_list_to_ap

(eval_list, total_gt, threshold, method='score_sort')

lib/dataset/panoptic.py:711

↓ 1 callersMethod_eval_list_to_mpjpe

(eval_list, threshold=500)

lib/dataset/h36m.py:314

↓ 1 callersMethod_eval_list_to_mpjpe

(eval_list, threshold=500, method='score_sort')

lib/dataset/panoptic.py:743

↓ 1 callersMethod_eval_list_to_recall

(eval_list, total_gt, threshold=500)

lib/dataset/h36m.py:327

↓ 1 callersMethod_eval_list_to_recall

(eval_list, total_gt, threshold=500)

lib/dataset/panoptic.py:761

↓ 1 callersMethod_get_cam

(self)

lib/dataset/shelf.py:233

↓ 1 callersMethod_get_cam

(self, camera)

lib/dataset/h36m.py:174

↓ 1 callersMethod_get_cam

(self)

lib/dataset/campus.py:228

↓ 1 callersMethod_get_cam

(self, seq)

lib/dataset/panoptic.py:454

↓ 1 callersFunction_get_clones

(module, N)

lib/models/mvp_decoder.py:328

↓ 1 callersMethod_get_db

(self, pesudo_gt_path)

lib/dataset/shelf.py:124

↓ 1 callersMethod_get_db

(self)

lib/dataset/h36m.py:103

↓ 1 callersMethod_get_db

(self, pesudo_gt_path)

lib/dataset/campus.py:111

↓ 1 callersMethod_get_db

(self)

lib/dataset/panoptic.py:303

↓ 1 callersMethod_get_deconv_cfg

(self, deconv_kernel, index)

lib/mvn/models/pose_resnet.py:253

↓ 1 callersMethod_get_deconv_cfg

(self, deconv_kernel, index)

lib/models/pose_resnet.py:158

↓ 1 callersMethod_get_group

(self, db)

lib/dataset/h36m.py:185

↓ 1 callersMethod_get_key_str

(self, datum)

lib/dataset/h36m.py:212

↓ 1 callersMethod_get_tgt_permutation_idx

(self, indices)

lib/models/multi_view_pose_transformer.py:783

↓ 1 callersMethod_initialize_weights

(self)

lib/mvn/models/v2v.py:171

↓ 1 callersFunction_is_power_of_2

(n)

lib/models/ops/modules/projattn.py:34

↓ 1 callersMethod_make_deconv_layer

(self, num_layers, num_filters, num_kernels)

lib/mvn/models/pose_resnet.py:266

↓ 1 callersMethod_make_deconv_layer

(self, num_layers, num_filters, num_kernels)

lib/models/pose_resnet.py:171

↓ 1 callersFunction_max_by_axis

(the_list)

lib/utils/misc.py:303

↓ 1 callersFunction_max_by_axis

(the_list)

lib/models/util/misc.py:390

↓ 1 callersMethod_read_configs

(self)

lib/smpl/smpl_common.py:191

↓ 1 callersFunction_rescore

Rescoring mechanism gaussian or linear. Args: overlap: calculated ious scores: target scores. thr: retain oks overlap < t

lib/core/nms.py:131

↓ 1 callersMethod_reset_parameters

(self)

lib/models/multi_view_pose_transformer.py:274

↓ 1 callersFunctionadd_path

(path)

process/_init_paths.py:28

↓ 1 callersFunctionaverage_bone_length

(loader, save_dir, prefix='bone_length', consider_n=300)

process/extract_dataset.py:186

↓ 1 callersMethodbayesian_update

(self, new_reference_points, reference_points_prior, confidences_filter)

lib/models/dq_decoder.py:1047

↓ 1 callersFunctionbox_iou

(boxes1, boxes2)

lib/models/util/box_ops.py:34

↓ 1 callersMethodbuild

(self)

lib/mvn/utils/volumetric.py:49

↓ 1 callersMethodcalculate_2d_offsets

From aggregated features, calculate 2d offsets for each projected 2D joints. TODO: Further speed up with batch processing.

lib/models/dq_decoder.py:659

↓ 1 callersFunctioncamera_array_to_name

(array)

lib/mvn/datasets/human36m_preprocessing/generate-labels-npy-multiview.py:67

↓ 1 callersFunctioncamera_to_world_frame

Args x: Nx3 points in camera coordinates R: 3x3 Camera rotation matrix T: 3x1 Camera translation parameters Returns

lib/utils/cameras_cpu.py:82

↓ 1 callersFunctioncompute_similarity_transform

A port of MATLAB's `procrustes` function to Numpy. Adapted from http://stackoverflow.com/a/18927641/1884420 Args X: array NxM of

lib/utils/transforms.py:234

↓ 1 callersFunctioncompute_similarity_transform

A port of MATLAB's `procrustes` function to PyTorch. Adapted from http://stackoverflow.com/a/18927641/1884420 Computes a similarity trans

lib/core/loss.py:299

↓ 1 callersMethodcreate_layers

(self)

lib/smpl/smpl_common.py:85

↓ 1 callersFunctioncrop_image

Crops area from image specified as bbox. Always returns area of size as bbox filling missing parts with zeros Args: image numpy array of s

lib/mvn/utils/img.py:10

↓ 1 callersFunctiondraw_voxels

(voxels, ax, shape=(8, 8, 8), norm=True, alpha=0.1)

lib/mvn/utils/vis.py:419

↓ 1 callersFunctionfreeze_defaultdict

(x)

lib/mvn/datasets/human36m_preprocessing/collect-bboxes.py:110

↓ 1 callersFunctiongaussian_2d_pdf

(coords, means, sigmas, normalize=True)

lib/mvn/utils/op.py:169

↓ 1 callersFunctiongen_config

(config_file)

lib/core/config.py:395

↓ 1 callersMethodgenerate_T_pose

roots: (query_num, 3)

lib/models/dq_transformer.py:226

↓ 1 callersMethodgenerate_T_pose_batch

roots: (batch, query_num, 3)

lib/models/dq_transformer.py:238

↓ 1 callersFunctiongenerate_batch_query_ids

(indices)

lib/models/dq_decoder.py:1212

↓ 1 callersMethodgenerate_features

Use projective attention to aggregate features from different views. For each projected joints, we sample attention points nearby to

lib/models/dq_decoder.py:516

↓ 1 callersFunctionget_2d_coords

(image_size, H, W, K, R, T, ret_rays_o=False)

lib/models/position_encoding.py:239

↓ 1 callersMethodget_conv_mat

Get the conversion matrix and its inverse.

lib/structural/structural_triangulation.py:42

↓ 1 callersFunctionget_extensions

()

lib/models/ops/setup.py:37

↓ 1 callersFunctionget_host_info

()

run/train_3d.py:80

↓ 1 callersFunctionget_inner_mat

(u, v)

lib/structural/structural_triangulation.py:181

↓ 1 callersFunctionget_model_name

(cfg)

lib/core/config.py:425

↓ 1 callersFunctionget_optimizer

(model_without_ddp, weight_decay, optim_type)

run/train_3d.py:116

↓ 1 callersFunctionget_rank

()

lib/utils/misc.py:385

↓ 1 callersFunctionget_rank

()

lib/models/util/misc.py:477

↓ 1 callersFunctionget_ray_directions

Get ray directions for all pixels in camera coordinate. Reference: https://www.scratchapixel.com/lessons/3d-basic-rendering/ r

lib/models/position_encoding.py:188

↓ 1 callersFunctionget_rays_new

(image_size, H, W, K, R, T, ret_rays_o=False)

lib/models/position_encoding.py:256

↓ 1 callersFunctionget_rotation_matrix

Returns the rotation matrix associated with counterclockwise rotation about the given axis by theta radians.

lib/mvn/utils/volumetric.py:87

↓ 1 callersFunctionget_scale

(image_size, resized_size)

lib/utils/transforms.py:170

↓ 1 callersFunctionget_total_grad_norm

(parameters, norm_type=2)

lib/models/util/misc.py:597

↓ 1 callersFunctionimage_batch_to_torch

(image_batch)

lib/mvn/utils/img.py:96

↓ 1 callersMethodinit_weights

(self, pretrained='')

lib/models/pose_resnet.py:218

↓ 1 callersMethodinitialize_reference_points

Averagely sample 3D queries coordinates to cover all the space as coarse initialization

lib/models/dq_transformer.py:250

↓ 1 callersFunctionis_main_process

()

lib/models/util/misc.py:495

↓ 1 callersFunctionjudge_float_or_int

(x)

lib/utils/string_parser.py:13

↓ 1 callersMethodlearnable_triangulate

Learnable triangulation to recover 3D poses from refined 2D poses, in a differentiable way. input: @keypoints_2d: 2d pos

lib/models/dq_decoder.py:399

↓ 1 callersFunctionload_backbone_panoptic

(model, pretrained_file)

lib/utils/utils.py:152

↓ 1 callersFunctionload_checkpoint

(model, optimizer, output_dir, filename='checkpoint.pth.tar')

lib/utils/utils.py:105

↓ 1 callersFunctionload_checkpoint_best

(model, optimizer, output_dir, filename='checkpoint.pth.tar')

lib/utils/utils.py:124

↓ 1 callersMethodlog_likelihood

Create graph operation for negative log-likelihood calculation

lib/smpl/prior.py:208

← previous next →101–200 of 561, ranked by callers