MCPcopy Create free account

hub / github.com/EGO4D/episodic-memory / functions

Functions1,142 in github.com/EGO4D/episodic-memory

↓ 2 callersFunctionComputePointJacobian
Compute the point Jacobian Parameters ---------- X : ndarray of shape (3,) 3D point p : ndarray of shape (7,) Ca
VQ3D/camera_pose_estimation/reconstruction.py:170
↓ 2 callersFunctionMLP
Multi-layer perceptron
VQ3D/camera_pose_estimation/SuperGlueMatching/models/superglue.py:49
↓ 2 callersFunctionPnP
(x1s, f3ds, x2s, m1_ids, K1, K2, thres=1e-2)
VQ3D/camera_pose_estimation/pnp_api.py:12
↓ 2 callersMethod__init__
(self, dataset_name='walter_basement_03', resize=[640, 480], root='', start_idx=0, end_idx=10
VQ3D/camera_pose_estimation/SuperGlueMatching/extract_visual_database.py:60
↓ 2 callersMethod__init__
(self, input_dim, hidden_dim, output_dim, num_layers, BN=False)
EgoTracks/tracking/models/stark_tracker/head.py:224
↓ 2 callersMethod__init__
( self, num_pos_feats=64, temperature=10000, normalize=False, scale=None )
EgoTracks/tracking/models/stark_tracker/position_encoding.py:17
↓ 2 callersMethod__init__
( self, in_features_1: int, in_features_2: int, hidden_size: int, test
VQ2D/detectron2_extensions/modeling/roi_heads/siam_heads.py:79
↓ 2 callersMethod__len__
Gets the number of templates Args: None Returns: int: the length of the templates list
EgoTracks/tracking/models/template.py:120
↓ 2 callersMethod_call_one_stage
(self, cls_pred, reg_pred, all_anchors)
MQ/Models/ActionGenerator.py:30
↓ 2 callersFunction_check_use_shared_memory
()
EgoTracks/tracking/dataset/dataloader.py:13
↓ 2 callersMethod_forward_siam
Forward logic of the siam prediction branch. If `self.train_on_pred_boxes is True`, the function puts predicted boxes in the `pro
VQ2D/detectron2_extensions/modeling/roi_heads/set_heads.py:800
↓ 2 callersMethod_forward_siam
Forward logic of the siam prediction branch. If `self.train_on_pred_boxes is True`, the function puts predicted boxes in the `pro
VQ2D/detectron2_extensions/modeling/roi_heads/siam_heads.py:472
↓ 2 callersFunction_get_box
(annot_box)
VQ3D/VQ3D/scripts/run.py:20
↓ 2 callersFunction_get_box
(annot_box)
VQ2D/visualizations/visualize_annotations.py:11
↓ 2 callersMethod_get_class
(self, seq_id)
EgoTracks/tracking/dataset/train_datasets/tracking_net.py:170
↓ 2 callersMethod_get_class
(self, seq_path)
EgoTracks/tracking/dataset/train_datasets/lasot.py:163
↓ 2 callersMethod_get_sequence_path
(self, seq_id)
EgoTracks/tracking/dataset/train_datasets/got10k.py:200
↓ 2 callersMethod_get_video_data
(self, index)
MQ/Utils/dataset.py:81
↓ 2 callersMethod_ioa_with_anchors
(self, anchors_min, anchors_max, box_min, box_max)
MQ/Utils/dataset.py:180
↓ 2 callersMethod_loss_one_stage
(self, cls_pred, reg_pred, gt_bbox, num_gt, anchors, stage=0)
MQ/Models/Loss.py:59
↓ 2 callersMethod_losses
Standard binary cross entropy loss between predictions and ground-truth boxes. Args: same as :meth: `losses()` Returns:
VQ2D/detectron2_extensions/modeling/roi_heads/set_heads.py:496
↓ 2 callersMethod_losses
Standard binary cross entropy loss between predictions and ground-truth boxes. Args: same as :meth: `losses()` Returns:
VQ2D/detectron2_extensions/modeling/roi_heads/siam_heads.py:224
↓ 2 callersMethod_make_levels
(self, in_channels, out_channels)
MQ/Models/XGPN.py:62
↓ 2 callersFunctionapprox_equal_durations
(dur1, dur2, thresh=1.0)
VQ2D/convert_videos_to_clips.py:55
↓ 2 callersFunctionarange_like
(x, dim: int)
VQ3D/camera_pose_estimation/SuperGlueMatching/models/superglue.py:173
↓ 2 callersFunctionbuild_augmentation
Create a list of default :class:`Augmentation` from config. Now it includes resizing and flipping. Returns: list[Augmentation]
VQ2D/vq2d/baselines/dataloader.py:16
↓ 2 callersFunctionbuild_box_head
(cfg)
EgoTracks/tracking/models/stark_tracker/head.py:244
↓ 2 callersFunctionbuild_optimizer
(cfg, model)
EgoTracks/tracking/solver/build.py:14
↓ 2 callersFunctionbuild_starkst
(cfg)
EgoTracks/tracking/models/stark_tracker/stark_st.py:92
↓ 2 callersFunctionbuild_transformer
(cfg)
EgoTracks/tracking/models/stark_tracker/transformer.py:589
↓ 2 callersFunctioncenter2corner
:param center: Center or np.array 4*N :return: Corner or np.array 4*N
EgoTracks/tracking/utils/bbox_helper.py:33
↓ 2 callersFunctionclip_box
(box: List, H, W, margin=0)
EgoTracks/tracking/models/stark_tracker/utils/box_ops.py:96
↓ 2 callersMethodcompute_loss
(self, start_logits, end_logits, start_labels, end_labels)
NLQ/VSLNet/model/VSLNet.py:145
↓ 2 callersFunctioncompute_visual_query_metrics
Compute model performance on the visual query task. Includes the following metrics: * Temporal AP * SpatioTemporal AP * S
VQ2D/vq2d/metrics/metrics.py:21
↓ 2 callersFunctionconv3x3
(in_channels: int, out_channels: int, stride: int = 1)
VQ2D/detectron2_extensions/modeling/roi_heads/set_heads.py:47
↓ 2 callersFunctionconv3x3
(in_channels: int, out_channels: int, stride: int = 1)
VQ2D/detectron2_extensions/modeling/roi_heads/siam_heads.py:38
↓ 2 callersFunctionconvert_image_np2torch
Converts an array of images from numpy to pytorch after normalization. Parameters: ----------- image: (B, H, W, 3) numpy array
VQ2D/vq2d/tracking/pfilter.py:160
↓ 2 callersFunctionconvert_length_to_mask
(lengths)
NLQ/VSLNet/utils/runner_utils.py:53
↓ 2 callersFunctioncreate_indices
(positions, weights)
VQ2D/vq2d/tracking/pfilter.py:52
↓ 2 callersFunctiondefault_argument_parser
Create a parser with some common arguments used by tracking. Args: epilog (str): epilog passed to ArgumentParser describing the usag
EgoTracks/tracking/utils/defaults.py:19
↓ 2 callersFunctiondisplay_results
(eval_result, miou, title=None)
NLQ/2D-TAN/lib/core/eval.py:108
↓ 2 callersFunctiondraw_box_on_image
(image, box, color=(255, 0, 0), thickness=5)
VQ2D/visualizations/visualize_annotations.py:37
↓ 2 callersFunctioneval
(segments, data)
NLQ/2D-TAN/lib/core/eval.py:55
↓ 2 callersFunctioneval_main
(args, cfg=None)
EgoTracks/tools/eval_net.py:40
↓ 2 callersFunctioneval_test
( model, data_loader, device, mode="test", result_save_path=None, gt_json_path=None,
NLQ/VSLNet/utils/runner_utils.py:62
↓ 2 callersMethodforward_box_head
hs: output embeddings (1, B, N, C) memory: encoder embeddings (HW1+HW2, B, C)
EgoTracks/tracking/models/stark_tracker/stark_s.py:95
↓ 2 callersMethodforward_pass
(self, data, run_box_head, run_cls_head)
EgoTracks/tools/trainers/starkst_trainer.py:90
↓ 2 callersFunctionget_cfg
Get a copy of the default config. Returns: a detectron2 CfgNode instance.
VQ2D/detectron2_extensions/config/defaults.py:69
↓ 2 callersMethodget_frame_ids_stark
(self, visible, valid)
EgoTracks/tracking/dataset/trackingdataset.py:442
↓ 2 callersMethodget_frame_ids_trident
(self, visible)
EgoTracks/tracking/dataset/trackingdataset.py:404
↓ 2 callersMethodget_name
(self)
EgoTracks/tracking/dataset/train_datasets/lasot.py:95
↓ 2 callersFunctionget_neigh_idx_semantic
(x, n_neigh, num_frms, opt)
MQ/Models/GCNs.py:38
↓ 2 callersMethodget_one_search
(self)
EgoTracks/tracking/dataset/trackingdataset.py:376
↓ 2 callersMethodget_sequence_info
(self, seq_id)
EgoTracks/tracking/dataset/train_datasets/lasot.py:146
↓ 2 callersFunctionget_test_loader
(dataset, video_features, configs)
NLQ/VSLNet/utils/data_loader.py:127
↓ 2 callersFunctionindependent_sample
Take a list of functions that each draw n samples from a distribution and concatenate the result into an n, d matrix Parameters: ---------
VQ2D/vq2d/tracking/pfilter.py:140
↓ 2 callersMethodinference
Run tracking model on a video sequence. Args: video: either a torch Tensor that contains a sequence of images, [N, 3, H,
EgoTracks/tracking/models/tracker.py:32
↓ 2 callersMethodinit_filter
Initialise the filter by drawing samples from the prior. Parameters: ----------- mask : array, optional boolean m
VQ2D/vq2d/tracking/pfilter.py:310
↓ 2 callersMethodinit_parameters
(self)
NLQ/VSLNet/model/VSLNet.py:103
↓ 2 callersMethodinit_tracker
This function is used to initilize SOT with first frame annotation Args: image: this is one image where we used to initi
EgoTracks/tracking/models/single_object_tracker.py:231
↓ 2 callersMethodinitialize_tracker
(self, frame, box)
VQ2D/vq2d/tracking/kys.py:35
↓ 2 callersFunctioninterpolated_prec_rec
Interpolated AP - VOCdevkit from VOC 2011.
MQ/Evaluation/utils.py:14
↓ 2 callersFunctioninterpolated_prec_rec
Interpolated AP - VOCdevkit from VOC 2011.
VQ2D/vq2d/metrics/utils.py:43
↓ 2 callersFunctionis_dist_avail_and_initialized
()
EgoTracks/tracking/models/stark_tracker/utils/misc.py:13
↓ 2 callersFunctionkl_div
Same as `torch.nn.functional.kl_div`, but returns 0 (instead of nan) for empty inputs. `flip_class` inverts the target since the positive cla
VQ2D/detectron2_extensions/layers/wrappers.py:53
↓ 2 callersFunctionlaunch_job
Run 'func' on one or more GPUs, specified in cfg Args: cfg (NamedTuple): configs. Details can be found in types.py init_metho
EgoTracks/tracking/utils/multiprocessing.py:68
↓ 2 callersMethodload_pose
(self, dirname: str)
VQ3D/VQ3D/API/get_query_3d_ground_truth.py:18
↓ 2 callersFunctionmake_matching_plot
(image0, image1, kpts0, kpts1, mkpts0, mkpts1, color, text, path, show_keypoints=False,
VQ3D/camera_pose_estimation/SuperGlueMatching/models/utils.py:452
↓ 2 callersMethodmap_box_back
(self, pred_box: List, resize_factor: float, pos: List = None)
EgoTracks/tracking/models/stark_tracker/stark_tracker.py:438
↓ 2 callersFunctionnames2datasets
(name_list: List, cfg, image_loader=opencv_loader)
EgoTracks/tracking/dataset/build.py:21
↓ 2 callersFunctionnms
Pure Python NMS baseline.
NLQ/2D-TAN/moment_localization/test.py:77
↓ 2 callersFunctionnormalize_keypoints
Normalize keypoints locations based on image image_shape
VQ3D/camera_pose_estimation/SuperGlueMatching/models/superglue.py:63
↓ 2 callersFunctionpad_char_seq
(sequences, max_length=None, max_length_2=None)
NLQ/VSLNet/utils/data_util.py:149
↓ 2 callersFunctionpad_video_seq
(sequences, max_length=None)
NLQ/VSLNet/utils/data_util.py:166
↓ 2 callersFunctionplot_keypoints
(kpts0, kpts1, color='w', ps=2)
VQ3D/camera_pose_estimation/SuperGlueMatching/models/utils.py:429
↓ 2 callersMethodprepare_targets
(self, gt_bbox, num_gt, anchors, stage=0)
MQ/Models/Loss.py:132
↓ 2 callersMethodpreprocess_image
Normalize, pad and batch the input images.
VQ2D/detectron2_extensions/modeling/meta_arch/siam_rcnn.py:146
↓ 2 callersFunctionread_video_md
(path)
VQ2D/convert_videos_to_images.py:16
↓ 2 callersMethodreg_loss_func
(self, pred, target, anchor, pred_boxes=None, weight=None)
MQ/Models/Loss.py:97
↓ 2 callersMethodreset
(self)
VQ3D/camera_pose_estimation/SuperGlueMatching/models/utils.py:67
↓ 2 callersMethodreset
(self)
EgoTracks/tools/trainers/base_trainer.py:304
↓ 2 callersMethodreset_parameters
(self)
EgoTracks/tracking/models/stark_tracker/position_encoding.py:82
↓ 2 callersMethodreset_tracker
To reset the tracker after done running on one video. Clear any intermediate variables to get a clean setup for next video.
EgoTracks/tracking/models/single_object_tracker.py:243
↓ 2 callersFunctionrotate_intrinsics
image_shape is the shape of the image after rotation
VQ3D/camera_pose_estimation/SuperGlueMatching/models/utils.py:315
↓ 2 callersFunctionrotate_pose_inplane
(i_T_w, rot)
VQ3D/camera_pose_estimation/SuperGlueMatching/models/utils.py:335
↓ 2 callersMethodrun_model
Most tracking models run frame by frame. This function runs a single forward of the tracking model. Args: image: this is
EgoTracks/tracking/models/single_object_tracker.py:202
↓ 2 callersFunctionsample_target
Extracts a square crop centered at target_bb box, of area search_area_factor^2 times target_bb area args: im - cv image target_bb
EgoTracks/tracking/dataset/processing/processing_utils.py:13
↓ 2 callersFunctionscale_intrinsics
(K, scales)
VQ3D/camera_pose_estimation/SuperGlueMatching/models/utils.py:346
↓ 2 callersMethodselect_good_frames
r""" Selects good frames for intrinsics parameter estimation using COLMAP. A good frame is non-blury - it has a variance of Laplacian
VQ3D/camera_pose_estimation/Camera_Intrinsics_API/get_camera_intrinsics.py:103
↓ 2 callersFunctionskewsymm
(x)
VQ3D/camera_pose_estimation/utils.py:114
↓ 2 callersFunctiontime_to_index
(start_time, end_time, num_units, duration)
NLQ/VSLNet/utils/data_util.py:106
↓ 2 callersFunctionto_homogeneous
(points)
VQ3D/camera_pose_estimation/SuperGlueMatching/models/utils.py:351
↓ 2 callersMethodto_json
(self)
VQ2D/vq2d/structures.py:20
↓ 2 callersFunctiontrain_VSGN_epoch
(data_loader, model, optimizer, epoch, writer, opt, is_train=True)
MQ/Train.py:61
↓ 2 callersFunctiontriplet_margin
Returns 0 (instead of nan) for empty inputs. `flip_class` inverts the target since the positive class is 0 and the negative/background class
VQ2D/detectron2_extensions/layers/wrappers.py:79
↓ 2 callersMethodupdate_bd
(self, anchors, start_offsets, end_offsets)
MQ/Models/BoundaryAdjust.py:82
↓ 2 callersMethodupdate_state
(self, frame)
VQ2D/vq2d/tracking/kys.py:55
↓ 2 callersMethodupdate_tracker
This function is used to update tracking information as the model runs through the video. In SOT, the most important update is to upd
EgoTracks/tracking/models/single_object_tracker.py:217
↓ 2 callersFunctionvalid_2Dfeatures
(x)
VQ3D/camera_pose_estimation/sfm_api_wsuperglue.py:111
↓ 2 callersMethodvolume
(self)
VQ2D/vq2d/structures.py:89
← previousnext →101–200 of 1,142, ranked by callers