Types & classes — github.com/apple/ml-aspanformer

CodeHub MCPcopy Create free account

hub / github.com/apple/ml-aspanformer / types & classes

Types & classes30 in github.com/apple/ml-aspanformer

⨍ Functions 198◇ Types & classes 30

↓ 4 callersClasslayernorm2d

src/ASpanFormer/aspan_module/attention.py:7

↓ 2 callersClassASpanFormer

src/ASpanFormer/aspanformer.py:13

↓ 2 callersClassFullAttention

src/ASpanFormer/aspan_module/attention.py:134

↓ 2 callersClassMultiSceneDataModule

For distributed training, each training process is assgined only a part of the training scenes to reduce memory overhead.

src/lightning/data.py:31

↓ 2 callersClassPL_ASpanFormer

src/lightning/lightning_aspanformer.py:27

↓ 2 callersClassmessageLayer_gla

src/ASpanFormer/aspan_module/transformer.py:66

↓ 1 callersClassASpanLoss

src/losses/aspan_loss.py:6

↓ 1 callersClassCoarseMatching

src/ASpanFormer/utils/coarse_matching.py:61

↓ 1 callersClassDarkAug

Extreme dark augmentation aiming at Aachen Day-Night

src/utils/augment.py:4

↓ 1 callersClassFineMatching

FineMatching with s2d paradigm

src/ASpanFormer/utils/fine_matching.py:9

↓ 1 callersClassFinePreprocess

src/ASpanFormer/aspan_module/fine_preprocess.py:7

↓ 1 callersClassHierachicalAttention

src/ASpanFormer/aspan_module/attention.py:21

↓ 1 callersClassInferenceProfiler

This profiler records duration of actions with cuda.synchronize() Use this in test time.

src/utils/profiler.py:7

↓ 1 callersClassLinearAttention

src/ASpanFormer/aspan_module/attention.py:165

↓ 1 callersClassLoFTREncoderLayer

src/ASpanFormer/aspan_module/loftr.py:6

↓ 1 callersClassLocalFeatureTransformer

A Local Feature Transformer (LoFTR) module.

src/ASpanFormer/aspan_module/loftr.py:65

↓ 1 callersClassLocalFeatureTransformer_Flow

A Local Feature Transformer (LoFTR) module.

src/ASpanFormer/aspan_module/transformer.py:188

↓ 1 callersClassMegaDepthDataset

src/datasets/megadepth.py:11

↓ 1 callersClassMobileAug

Random augmentations aiming at images of mobile/handhold devices.

src/utils/augment.py:22

↓ 1 callersClassPositionEncodingSine

This is a sinusoidal position encoding that generalized to 2-dimensional images

src/ASpanFormer/utils/position_encoding.py:6

↓ 1 callersClassRGBDFrame

tools/SensorData.py:12

↓ 1 callersClassRandomConcatSampler

Random sampler for ConcatDataset. At each epoch, `n_samples_per_subset` samples will be draw from each subset in the ConcatDataset. If `subset_re

src/datasets/sampler.py:5

↓ 1 callersClassResNetFPN_16_4

ResNet+FPN, output resolution are 1/16 and 1/4. Each block has 2 layers.

src/ASpanFormer/backbone/resnet_fpn.py:121

↓ 1 callersClassResNetFPN_8_2

ResNet+FPN, output resolution are 1/8 and 1/2. Each block has 2 layers.

src/ASpanFormer/backbone/resnet_fpn.py:43

↓ 1 callersClassScanNetDataset

src/datasets/scannet.py:17

↓ 1 callersClassSensorData

tools/SensorData.py:46

↓ 1 callersClassflow_initializer

src/ASpanFormer/aspan_module/transformer.py:136

↓ 1 callersClassmessageLayer_ini

src/ASpanFormer/aspan_module/transformer.py:8

ClassBasicBlock

src/ASpanFormer/backbone/resnet_fpn.py:15

ClassTqdmBatchCompletionCallback

src/utils/misc.py:87