Functions — github.com/CHENGY12/DMML

CodeHub MCPcopy Create free account

hub / github.com/CHENGY12/DMML / functions

Functions54 in github.com/CHENGY12/DMML

⨍ Functions 54 ◇ Types & classes 10

↓ 4 callersFunctioneuclidean_dist

Compute (Squared) Euclidean distance between two tensors. Args: x: input tensor with size N x D. y: input tensor with size M

loss/common.py:4

↓ 4 callersFunctionget_mask

Generate positive and negative masks for contrastive and triplet loss.

loss/common.py:45

↓ 4 callersMethodid

file_path: unix style file path return: person id

data/duke.py:45

↓ 3 callersMethodid

file_path: unix style file path return: person id

data/market1501.py:46

↓ 2 callersFunctionextract_feature

Extract query and gallery features.

↓ 2 callersFunctionget_id

Get camera ids and class labels of images.

↓ 2 callersFunctionlist_pictures

(directory, ext='jpg|jpeg|bmp|png|ppm')

data/common.py:5

↓ 2 callersFunctionmain

↓ 1 callersMethod_sample

(population, k)

data/sampler.py:43

↓ 1 callersFunctionadjust_lr_exp

Adjust learning rate exponentially from a given epoch to the end of training.

↓ 1 callersMethodcamera

file_path: unix style file path return: camera id

data/duke.py:53

↓ 1 callersMethodcamera

file_path: unix style file path return: camera id

data/market1501.py:54

↓ 1 callersFunctioncompute_map

Compute mean average precision (mAP) and cumulative matching characteristic (CMC) scores.

↓ 1 callersFunctioncosine_dist

loss/common.py:30

↓ 1 callersFunctioneval

(gid, dataset, dataset_root, which, exp_dir)

↓ 1 callersFunctionevaluate

↓ 1 callersFunctionflip_img

Do horizontal flipping for a given image.

↓ 1 callersFunctionget_parser

↓ 1 callersFunctioninit_seed

Set random seed for torch and numpy.

↓ 1 callersFunctionmain

↓ 1 callersFunctionmake_dataloader

Make train dataloader. Args: epoch: current epoch number, used in random erasing data augmentation.

data/__init__.py:13

↓ 1 callersFunctionmake_loss

Construct loss function(s).

loss/__init__.py:9

↓ 1 callersFunctionmake_model

Initialize ResNet-50 model.

↓ 1 callersFunctiontrain

data/random_erasing.py:25

Method__getitem__

data/duke.py:31

Method__getitem__

data/market1501.py:32

(self, num_classes=None, include_top=False, remove_downsample=False)

(self, dataset_root, transform, split='train')

data/duke.py:13

(self, dataset_root, transform, split='train')

data/market1501.py:13

(self, data_source, batch_k)

data/sampler.py:11

(self, probability=0.5, s_epoch=0.1, ratio=0.3, mean=[0.4914, 0.4822, 0.4465])

data/random_erasing.py:18

(self, reg_lambda=0.002, gid=None)

loss/npair.py:13

loss/triplet.py:15

(self, num_support, distance_mode='hard_mining', margin=0.4, gid=None)

loss/dmml.py:18

loss/contrastive.py:14

(self, margin=0.4, gid=None)

loss/lifted.py:11

data/sampler.py:30

data/duke.py:41

data/market1501.py:42

data/sampler.py:39

return: camera id list corresponding to dataset image paths

data/duke.py:75

return: camera id list corresponding to dataset image paths

data/market1501.py:76

Functionfloat_or_string

Tries to convert the string to float, otherwise returns the string.

(self, feature, label)

loss/npair.py:18

(self, feature, label)

loss/triplet.py:21

(self, feature, label)

loss/dmml.py:31

(self, feature, label)

loss/contrastive.py:20

(self, features, labels)

loss/lifted.py:16

return: person id list corresponding to dataset image paths

data/duke.py:61

return: person id list corresponding to dataset image paths

data/market1501.py:62

Methodunique_ids

return: unique person ids in ascending order

data/duke.py:68

Methodunique_ids

return: unique person ids in ascending order

data/market1501.py:69