New Code

Author: KeyForce

.gitignore

@@ -0,0 +1,107 @@
+*.sh
+.vscode
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+/Domain-Adaptation/.idea
+*.npz
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+.hypothesis/
+.pytest_cache/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# pyenv
+.python-version
+
+# celery beat schedule file
+celerybeat-schedule
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/

MGN/.gitignore

@@ -0,0 +1,11 @@
+.project
+.settings/
+.prefs
+.pydevproject
+.idea/
+.idea
+.DS_Store
+.cache
+*.pyc
+*.html
+*.xlm

MGN/README.md

@@ -0,0 +1,92 @@
+# Multiple Granularity Network
+Implement of paper:[Learning Discriminative Features with Multiple Granularities for Person Re-Identification](https://arxiv.org/abs/1804.01438v1)
+
+## Dependencies
+
+- Python >= 3.5
+- PyTorch >= 0.4.0
+- torchvision
+- scipy
+- numpy
+- scikit_learn
+
+
+
+## Current Result
+
+| Re-Ranking| backbone |  mAP | rank1 | rank3 | rank5 | rank10 |  
+| :------: | :------: |  :------: | :------: | :------: | :------: |  :------: |   
+| yes | resnet50 |  94.33 | 95.58 | 97.54 | 97.92 | 98.46 | 
+| no | resnet50 |  86.15 | 94.95 | 97.42 | 98.07 | 98.93 | 
+
+
+
+## Data
+
+The data structure would look like:
+```
+data/
+    bounding_box_train/
+    bounding_box_test/
+    query/
+```
+#### Market1501 
+Download from [here](http://www.liangzheng.org/Project/project_reid.html)
+
+#### DukeMTMC-reID
+Download from [here](http://vision.cs.duke.edu/DukeMTMC/)
+
+#### CUHK03
+1. Download cuhk03 dataset from "http://www.ee.cuhk.edu.hk/~xgwang/CUHK_identification.html"
+2. Unzip the file and you will get the cuhk03_release dir include cuhk-03.mat
+3. Download "cuhk03_new_protocol_config_detected.mat" from "https://github.com/zhunzhong07/person-re-ranking/tree/master/evaluation/data/CUHK03"
+and put it with cuhk-03.mat. We need this new protocol to split the dataset.
+```
+python utils/transform_cuhk03.py --src <path/to/cuhk03_release> --dst <path/to/save>
+```
+
+NOTICE:You need to change num_classes in network depend on how many people in your train dataset! e.g. 751 in Market1501
+
+## Weights
+
+Pretrained weight download from [google drive](https://drive.google.com/open?id=16V7ZsflBbINHPjh_UVYGBVO6NuSxEMTi)
+or [baidu drive](https://pan.baidu.com/s/12AkumLX10hLx9vh_SQwdyw) password:mrl5
+## Train
+
+You can specify more parameters in opt.py
+
+```
+python main.py --mode train --data_path <path/to/Market-1501-v15.09.15> 
+```
+
+## Evaluate
+
+Use pretrained weight or your trained weight
+
+```
+python main.py --mode evaluate --data_path <path/to/Market-1501-v15.09.15> --weight <path/to/weight_name.pt> 
+```
+
+## Visualize
+
+Visualize rank10 query result of one image(query from bounding_box_test)
+
+Extract features will take a few munutes, or you can save features as .mat file for multiple uses
+
+![image](https://s1.ax1x.com/2018/11/27/FV9xyj.png)
+
+```
+python main.py --mode vis --query_image <path/to/query_image> --weight <path/to/weight_name.pt> 
+```
+
+
+## Citation
+
+```text
+@ARTICLE{2018arXiv180401438W,
+    author = {{Wang}, G. and {Yuan}, Y. and {Chen}, X. and {Li}, J. and {Zhou}, X.},
+    title = "{Learning Discriminative Features with Multiple Granularities for Person Re-Identification}",
+    journal = {ArXiv e-prints},
+    year = 2018,
+}
+```

MGN/__init__.py

@@ -0,0 +1,8 @@
+# -*- coding: utf-8 -*-
+"""
+@File    : __init__.py
+@Time    : 2020/4/18 16:35
+@Author  : KeyForce
+@Email   : [email protected]
+"""
+__all__ = ['utils', 'data', 'loss', 'network']

MGN/data.py

@@ -0,0 +1,120 @@
+from torchvision import transforms
+from torch.utils.data import dataset, dataloader
+from torchvision.datasets.folder import default_loader
+from utils.RandomErasing import RandomErasing
+from utils.RandomSampler import RandomSampler
+from opt import opt
+import os
+import re
+
+
+class Data():
+    def __init__(self):
+        train_transform = transforms.Compose([
+            transforms.Resize((384, 128), interpolation=3),
+            transforms.RandomHorizontalFlip(),
+            transforms.ToTensor(),
+            transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
+            RandomErasing(probability=0.5, mean=[0.0, 0.0, 0.0])
+        ])
+
+        test_transform = transforms.Compose([
+            transforms.Resize((384, 128), interpolation=3),
+            transforms.ToTensor(),
+            transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
+        ])
+
+        self.trainset = Market1501(train_transform, 'train', opt.data_path)
+        self.testset = Market1501(test_transform, 'test', opt.data_path)
+        self.queryset = Market1501(test_transform, 'query', opt.data_path)
+
+        self.train_loader = dataloader.DataLoader(self.trainset,
+                                                  sampler=RandomSampler(self.trainset, batch_id=opt.batchid,
+                                                                        batch_image=opt.batchimage),
+                                                  batch_size=opt.batchid * opt.batchimage, num_workers=8,
+                                                  pin_memory=True)
+        self.test_loader = dataloader.DataLoader(self.testset, batch_size=opt.batchtest, num_workers=8, pin_memory=True)
+        self.query_loader = dataloader.DataLoader(self.queryset, batch_size=opt.batchtest, num_workers=8,
+                                                  pin_memory=True)
+
+        if opt.mode == 'vis':
+            self.query_image = test_transform(default_loader(opt.query_image))
+
+
+class Market1501(dataset.Dataset):
+    def __init__(self, transform, dtype, data_path):
+
+        self.transform = transform
+        self.loader = default_loader
+        self.data_path = data_path
+
+        if dtype == 'train':
+            self.data_path += '/bounding_box_train'
+            # self.data_path += ''
+        elif dtype == 'test':
+            self.data_path += '/bounding_box_test'
+            # self.data_path += ''
+        else:
+            self.data_path += '/query'
+
+        self.imgs = [path for path in self.list_pictures(self.data_path) if self.id(path) != -1]
+
+        self._id2label = {_id: idx for idx, _id in enumerate(self.unique_ids)}
+
+    def __getitem__(self, index):
+        path = self.imgs[index]
+        target = self._id2label[self.id(path)]
+
+        img = self.loader(path)
+        if self.transform is not None:
+            img = self.transform(img)
+
+        return img, target
+
+    def __len__(self):
+        return len(self.imgs)
+
+    @staticmethod
+    def id(file_path):
+        """
+        :param file_path: unix style file path
+        :return: person id
+        """
+        return int(file_path.split('/')[-1].split('_')[0])
+
+    @staticmethod
+    def camera(file_path):
+        """
+        :param file_path: unix style file path
+        :return: camera id
+        """
+        return int(file_path.split('/')[-1].split('_')[1][1])
+
+    @property
+    def ids(self):
+        """
+        :return: person id list corresponding to dataset image paths
+        """
+        return [self.id(path) for path in self.imgs]
+
+    @property
+    def unique_ids(self):
+        """
+        :return: unique person ids in ascending order
+        """
+        return sorted(set(self.ids))
+
+    @property
+    def cameras(self):
+        """
+        :return: camera id list corresponding to dataset image paths
+        """
+        return [self.camera(path) for path in self.imgs]
+
+    @staticmethod
+    def list_pictures(directory, ext='jpg|jpeg|bmp|png|ppm|npy'):
+        assert os.path.isdir(directory), 'dataset is not exists!{}'.format(directory)
+
+        return sorted([os.path.join(root, f)
+                       for root, _, files in os.walk(directory) for f in files
+                       if re.match(r'([\w]+\.(?:' + ext + '))', f)])

MGN/loss.py

@@ -0,0 +1,27 @@
+from torch.nn import CrossEntropyLoss
+from torch.nn.modules import loss
+from utils.TripletLoss import TripletLoss
+
+
+class Loss(loss._Loss):
+    def __init__(self):
+        super(Loss, self).__init__()
+
+    def forward(self, outputs, labels):
+        cross_entropy_loss = CrossEntropyLoss()
+        triplet_loss = TripletLoss(margin=1.2)
+
+        Triplet_Loss = [triplet_loss(output, labels) for output in outputs[1:4]]
+        Triplet_Loss = sum(Triplet_Loss) / len(Triplet_Loss)
+
+        CrossEntropy_Loss = [cross_entropy_loss(output, labels) for output in outputs[4:]]
+        CrossEntropy_Loss = sum(CrossEntropy_Loss) / len(CrossEntropy_Loss)
+
+        loss_sum = Triplet_Loss + 2 * CrossEntropy_Loss
+
+        print('\rtotal loss:%.2f  Triplet_Loss:%.2f  CrossEntropy_Loss:%.2f' % (
+            loss_sum.data.cpu().numpy(),
+            Triplet_Loss.data.cpu().numpy(),
+            CrossEntropy_Loss.data.cpu().numpy()),
+              end=' ')
+        return loss_sum