Functions — github.com/apple/ml-aspanformer

CodeHub MCPcopy Create free account

hub / github.com/apple/ml-aspanformer / functions

Functions198 in github.com/apple/ml-aspanformer

⨍ Functions 198◇ Types & classes 30

↓ 1 callersFunctionmask_border

Mask borders with value Args: m (torch.Tensor): [N, H0, W0, H1, W1] b (int) v (m.dtype)

src/ASpanFormer/utils/coarse_matching.py:10

↓ 1 callersFunctionmask_border_with_padding

(m, bd, v, p_m0, p_m1)

src/ASpanFormer/utils/coarse_matching.py:30

↓ 1 callersFunctionparse_args

↓ 1 callersFunctionparse_args

↓ 1 callersMethodpartition_token

(self,q,k,v,offset,span_scale,maskv)

src/ASpanFormer/aspan_module/attention.py:91

↓ 1 callersFunctionread_scannet_pose

Read ScanNet's Camera2World pose and transform it to World2Camera. Returns: pose_w2c (np.ndarray): (4, 4)

src/utils/dataset.py:170

↓ 1 callersFunctionrelative_pose_error

(T_0to1, R, t, ignore_gt_t_thr=0.0)

src/utils/metrics.py:12

↓ 1 callersMethodresize_input

(self,data,train_res,df=32)

src/ASpanFormer/aspanformer.py:113

↓ 1 callersFunctionsetup_gpus

A temporary fix for pytorch-lighting 1.3.x

src/utils/misc.py:44

↓ 1 callersFunctionspvs_coarse

Update: data (dict): { "conf_matrix_gt": [N, hw0, hw1], 'spv_b_ids': [M] 'spv_i_ids': [M]

src/ASpanFormer/utils/supervision.py:22

↓ 1 callersFunctionspvs_fine

Update: data (dict):{ "expec_f_gt": [M, 2]}

src/ASpanFormer/utils/supervision.py:124

↓ 1 callersFunctiontqdm_joblib

Context manager to patch joblib to report into tqdm progress bar given as argument Usage: with tqdm_joblib(tqdm(desc="My calculation"

src/utils/misc.py:72

(self, *args, **kwargs)

src/utils/misc.py:91

src/utils/augment.py:18

src/utils/augment.py:37

Method__getitem__

src/datasets/scannet.py:68

Method__getitem__

src/datasets/megadepth.py:67

(self, filename)

tools/SensorData.py:48

(self, *args, **kwargs)

src/utils/misc.py:88

src/utils/profiler.py:13

src/utils/augment.py:9

src/utils/augment.py:27

src/ASpanFormer/aspanformer.py:14

src/ASpanFormer/backbone/resnet_fpn.py:49

src/ASpanFormer/backbone/resnet_fpn.py:127

src/ASpanFormer/utils/fine_matching.py:12

Args: max_shape (tuple): for 1/8 featmap, the max length of 256 corresponds to 2048 pixels temp_bug_fix (bool): As no

src/ASpanFormer/utils/position_encoding.py:11

src/ASpanFormer/utils/coarse_matching.py:62

src/ASpanFormer/aspan_module/loftr.py:68

(self,d_model,nhead,nsample,radius_scale,nlevel=3)

src/ASpanFormer/aspan_module/attention.py:22

(self,d_model,nhead)

src/ASpanFormer/aspan_module/attention.py:135

(self, eps=1e-6)

src/ASpanFormer/aspan_module/attention.py:166

src/ASpanFormer/aspan_module/fine_preprocess.py:8

(self,d_model,d_flow,d_value, nhead,radius_scale,nsample,update_flow=True)

src/ASpanFormer/aspan_module/transformer.py:68

(self, dim, dim_flow, nhead, layer_num)

src/ASpanFormer/aspan_module/transformer.py:138

src/ASpanFormer/aspan_module/transformer.py:191

Manage one scene of ScanNet Dataset. Args: root_dir (str): ScanNet root directory that contains scene folders. npz_pat

src/datasets/scannet.py:18

Manage one scene(npz_path) of MegaDepth dataset. Args: root_dir (str): megadepth root directory that has `phoeni

src/datasets/megadepth.py:12

(self, data_source: ConcatDataset, n_samples_per_subset: int,

src/datasets/sampler.py:21

(self, args, config)

src/lightning/data.py:36

TODO: - use the new version of PL logging API.

src/lightning/lightning_aspanformer.py:28

src/losses/aspan_loss.py:7

src/datasets/sampler.py:44

src/datasets/scannet.py:53

src/datasets/megadepth.py:64

src/datasets/sampler.py:41

Method_build_concat_dataset

( self, data_root, npz_names, npz_dir, intrinsic_path, mode,

src/lightning/data.py:192

Method_build_concat_dataset_parallel

( self, data_root, npz_names, npz_dir, intrinsic_path, mode,

src/lightning/data.py:242

Functioncompute_symmetrical_epipolar_errors_offset

Update: data (dict):{"epi_errs": [M]}

src/utils/metrics.py:71

Methodconfigure_optimizers

src/lightning/lightning_aspanformer.py:57

Functiondraw_match

(img1, img2, corr1, corr2,inlier=[True],color=None,radius1=1,radius2=1,resize=None)

src/utils/misc.py:111

Functiondraw_match

(img1, img2, corr1, corr2,inlier=[True],color=None,radius1=1,radius2=1,resize=None)

demo/demo_utils.py:16

Functiondraw_points

(img,points,color=(0,255,0),radius=3)

src/utils/misc.py:104

Functiondraw_points

(img,points,color=(0,255,0),radius=3)

demo/demo_utils.py:9

Functionelu_feature_map

src/ASpanFormer/aspan_module/attention.py:162

Functionextract

(seq,scannet_dir,split,dump_dir)

tools/extract.py:12

src/ASpanFormer/backbone/resnet_fpn.py:32

src/ASpanFormer/backbone/resnet_fpn.py:100

src/ASpanFormer/backbone/resnet_fpn.py:180

Args: feat0 (torch.Tensor): [M, WW, C] feat1 (torch.Tensor): [M, WW, C] data (dict) Update:

src/ASpanFormer/utils/fine_matching.py:15

Args: x: [N, C, H, W]

src/ASpanFormer/utils/position_encoding.py:42

Args: feat0 (torch.Tensor): [N, L, C] feat1 (torch.Tensor): [N, S, C] offset: [layer, B, H, W, 4] (*2)

src/ASpanFormer/utils/coarse_matching.py:89

Args: x (torch.Tensor): [N, L, C] source (torch.Tensor): [N, S, C] x_mask (torch.Tensor): [N, L] (optiona

src/ASpanFormer/aspan_module/loftr.py:34

Args: feat0 (torch.Tensor): [N, L, C] feat1 (torch.Tensor): [N, S, C] mask0 (torch.Tensor): [N, L] (optio

src/ASpanFormer/aspan_module/loftr.py:86

src/ASpanFormer/aspan_module/attention.py:15

Args: q,k,v (torch.Tensor): [B, C, L] mask (torch.Tensor): [B, L] flow (torch.Tensor): [B, H, W, 4]

src/ASpanFormer/aspan_module/attention.py:39

Multi-head scaled dot-product attention, a.k.a full attention. Args: q,k,v: [N, D, L] mask: [N, L] Returns:

src/ASpanFormer/aspan_module/attention.py:140

Multi-Head linear attention proposed in "Transformers are RNNs" Args: queries: [N, L, H, D] keys: [N, S, H, D]

src/ASpanFormer/aspan_module/attention.py:171

(self, feat_f0, feat_f1, feat_c0, feat_c1, data)

src/ASpanFormer/aspan_module/fine_preprocess.py:29

(self, x0, x1,pos0,pos1,mask0=None,mask1=None)

src/ASpanFormer/aspan_module/transformer.py:35

Args: x0 (torch.Tensor): [B, C, H, W] x1 (torch.Tensor): [B, C, H, W] flow_feature0 (torch.Tensor): [B, C

src/ASpanFormer/aspan_module/transformer.py:96

(self, feat0, feat1,pos0,pos1,mask0=None,mask1=None,ds0=[4,4],ds1=[4,4])

src/ASpanFormer/aspan_module/transformer.py:152

Args: feat0 (torch.Tensor): [N, C, H, W] feat1 (torch.Tensor): [N, C, H, W] pos1,pos2: [N, C, H, W]

src/ASpanFormer/aspan_module/transformer.py:215

Update: data (dict): update{ 'loss': [1] the reduced loss across a batch, 'loss_scalars' (dict):

src/losses/aspan_loss.py:185

Functionget_local_rank

Returns: The rank of the current process within the local (per-machine) process group.

src/utils/comm.py:38

Functionget_local_size

Returns: The size of the per-machine process group, i.e. the number of processes per machine.

src/utils/comm.py:51

Functionis_main_process

src/utils/comm.py:64

(condition, message, level)

src/utils/misc.py:26

Methodoptimizer_step

( self, epoch, batch_idx, optimizer, optimizer_idx, optimizer_closure, on_tpu, using_n

src/lightning/lightning_aspanformer.py:63

Functionread_gl3d_depth

src/utils/dataset.py:193

Functionread_gl3d_gray

src/utils/dataset.py:188

Functionread_scannet_intrinsic

Read ScanNet's intrinsic matrix and return the 3x3 matrix.

src/utils/dataset.py:181

Functionreduce_dict

Reduce the values in the dictionary from all processes so that process with rank 0 has the reduced results. Args: input_dict (di

src/utils/comm.py:236

(image,long_dim)

demo/demo_utils.py:4

Setup train / val / test dataset. This method will be called by PL automatically. Args: stage (str): 'fit' in training ph

src/lightning/data.py:106

Functionshared_random_seed

Returns: int: a random number that is the same across all workers. If workers need a shared RNG, they can use this shared see

src/utils/comm.py:222

Functionstr2bool

Functionsynchronize

Helper function to synchronize (barrier) among all processes when using distributed training

src/utils/comm.py:68

Methodtest_dataloader

(self, *args, **kwargs)

src/lightning/data.py:319

Methodtest_epoch_end

(self, outputs)

src/lightning/lightning_aspanformer.py:259

Methodtest_step

(self, batch, batch_idx)

src/lightning/lightning_aspanformer.py:232

Methodtrain_dataloader

Build training dataloader for ScanNet / MegaDepth.

src/lightning/data.py:292

Methodtraining_epoch_end

(self, outputs)

src/lightning/lightning_aspanformer.py:160

Methodtraining_step

(self, batch, batch_idx)

src/lightning/lightning_aspanformer.py:118

Functionupper_config

src/utils/misc.py:20

Methodval_dataloader

Build validation dataloader for ScanNet / MegaDepth.

src/lightning/data.py:306

Methodvalidation_epoch_end

(self, outputs)

src/lightning/lightning_aspanformer.py:185

Methodvalidation_step

(self, batch, batch_idx)

src/lightning/lightning_aspanformer.py:167

← previous101–198 of 198, ranked by callers