linting code

.flake8

@@ -1,4 +1,18 @@
 [flake8]
-ignore = E261, E501, W293
+ignore =
+    # At least two spaces before inline comment
+    E261,
+    # Line lengths are recommended to be no greater than 79 characters
+    E501,
+    # Missing whitespace around arithmetic operator 
+    E226,
+    # Blank line contains whitespace
+    W293,
+    # Do not use bare 'except'
+    E722,
+    # Line break after binary operator
+    W504,
+    # isort found an import in the wrong position
+    I001
 max-line-length = 79
-exclude = __init__.py, build
+exclude = __init__.py, build, torchreid/metrics/rank_cylib/

docs/conf.py

@@ -57,7 +57,7 @@ templates_path = ['_templates']
 # You can specify multiple suffix as a list of string:
 #
 source_suffix = ['.rst', '.md']
-#source_suffix = '.rst'
+# source_suffix = '.rst'
 source_parsers = {'.md': 'recommonmark.parser.CommonMarkParser'}
 
 # The master toctree document.

linter.sh

@@ -1,3 +1,11 @@
+echo "Running isort"
 isort -y -sp .
+echo "Done"
 
-yapf -i -r -vv . -e build
+echo "Running yapf"
+yapf -i -r -vv -e build .
+echo "Done"
+
+echo "Running flake8"
+flake8 .
+echo "Done"

projects/DML/dml.py

@@ -1,10 +1,7 @@
 from __future__ import division, print_function, absolute_import
-import time
-import datetime
 import torch
 from torch.nn import functional as F
 
-from torchreid import metrics
 from torchreid.utils import open_all_layers, open_specified_layers
 from torchreid.engine import Engine
 from torchreid.losses import TripletLoss, CrossEntropyLoss

projects/DML/main.py

@@ -1,4 +1,3 @@
-import os
 import sys
 import copy
 import time
@@ -15,8 +14,8 @@ from torchreid.utils import (
 
 from dml import ImageDMLEngine
 from default_config import (
-    imagedata_kwargs, optimizer_kwargs, videodata_kwargs, engine_run_kwargs,
-    get_default_config, lr_scheduler_kwargs
+    imagedata_kwargs, optimizer_kwargs, engine_run_kwargs, get_default_config,
+    lr_scheduler_kwargs
 )
 
 

projects/OSNet_AIN/osnet_search.py

@@ -10,6 +10,24 @@ NORM_AFFINE = False # enable affine transformations for normalization layer
 ##########
 # Basic layers
 ##########
+class IBN(nn.Module):
+    """Instance + Batch Normalization."""
+
+    def __init__(self, num_channels):
+        super(IBN, self).__init__()
+        half1 = int(num_channels / 2)
+        self.half = half1
+        half2 = num_channels - half1
+        self.IN = nn.InstanceNorm2d(half1, affine=NORM_AFFINE)
+        self.BN = nn.BatchNorm2d(half2, affine=NORM_AFFINE)
+
+    def forward(self, x):
+        split = torch.split(x, self.half, 1)
+        out1 = self.IN(split[0].contiguous())
+        out2 = self.BN(split[1].contiguous())
+        return torch.cat((out1, out2), 1)
+
+
 class ConvLayer(nn.Module):
     """Convolution layer (conv + bn + relu)."""
 

projects/OSNet_AIN/softmax_nas.py

@@ -1,11 +1,6 @@
 from __future__ import division, print_function, absolute_import
-import time
-import datetime
 
 from torchreid import metrics
-from torchreid.utils import (
-    AverageMeter, open_all_layers, open_specified_layers
-)
 from torchreid.engine import Engine
 from torchreid.losses import CrossEntropyLoss
 
@@ -58,7 +53,7 @@ class ImageSoftmaxNASEngine(Engine):
             lmda = self.init_lmda
         else:
             lmda = self.init_lmda * self.lmda_decay_rate**(
-                epoch // self.lmda_decay_step
+                self.epoch // self.lmda_decay_step
             )
             if lmda < self.min_lmda:
                 lmda = self.min_lmda

scripts/main.py

@@ -1,4 +1,3 @@
-import os
 import sys
 import time
 import os.path as osp

setup.py

@@ -45,8 +45,7 @@ def get_requirements(filename='requirements.txt'):
 setup(
     name='torchreid',
     version=find_version(),
-    description=
-    'A library for deep learning person re-identification in PyTorch',
+    description='A library for deep learning person re-ID in PyTorch',
     author='Kaiyang Zhou',
     license='MIT',
     long_description=readme(),

torchreid/data/datamanager.py

@@ -70,7 +70,7 @@ class DataManager(object):
         """Returns the number of training cameras."""
         return self._num_train_cams
 
-    def return_query_and_gallery_by_name(self, name):
+    def fetch_qg(self, name):
         """Returns query and gallery of a test dataset, each containing
         tuples of (img_path(s), pid, camid).
 

torchreid/data/datasets/dataset.py

@@ -238,9 +238,9 @@ class Dataset(object):
               '  gallery  | {:5d} | {:7d} | {:9d}\n' \
               '  ----------------------------------------\n' \
               '  items: images/tracklets for image/video dataset\n'.format(
-              num_train_pids, len(self.train), num_train_cams,
-              num_query_pids, len(self.query), num_query_cams,
-              num_gallery_pids, len(self.gallery), num_gallery_cams
+                  num_train_pids, len(self.train), num_train_cams,
+                  num_query_pids, len(self.query), num_query_cams,
+                  num_gallery_pids, len(self.gallery), num_gallery_cams
               )
 
         return msg

torchreid/data/datasets/image/cuhk03.py

@@ -151,10 +151,8 @@ class CUHK03(ImageDataset):
                     img_paths = _process_images(
                         camp[pid, :], campid, pid, imgs_dir
                     )
-                    assert len(img_paths
-                               ) > 0, 'campid{}-pid{} has no images'.format(
-                                   campid, pid
-                               )
+                    assert len(img_paths) > 0, \
+                        'campid{}-pid{} has no images'.format(campid, pid)
                     meta_data.append((campid + 1, pid + 1, img_paths))
                 print(
                     '- done camera pair {} with {} identities'.format(

torchreid/data/datasets/image/dukemtmcreid.py

@@ -61,7 +61,8 @@ class DukeMTMCreID(ImageDataset):
             pid, camid = map(int, pattern.search(img_path).groups())
             assert 1 <= camid <= 8
             camid -= 1 # index starts from 0
-            if relabel: pid = pid2label[pid]
+            if relabel:
+                pid = pid2label[pid]
             data.append((img_path, pid, camid))
 
         return data

torchreid/data/datasets/image/ilids.py

@@ -18,7 +18,7 @@ class iLIDS(ImageDataset):
     
     Dataset statistics:
         - identities: 119.
-        - images: 476. 
+        - images: 476.
         - cameras: 8 (not explicitly provided).
     """
     dataset_dir = 'ilids'

torchreid/data/datasets/image/msmt17.py

@@ -3,7 +3,7 @@ import os.path as osp
 
 from ..dataset import ImageDataset
 
-##### Log #####
+# Log
 # 22.01.2019
 # - add v2
 # - v1 and v2 differ in dir names

torchreid/data/datasets/image/viper.py

@@ -83,10 +83,8 @@ class VIPeR(ImageDataset):
                 np.random.shuffle(order)
                 train_idxs = order[:num_train_pids]
                 test_idxs = order[num_train_pids:]
-                assert not bool(
-                    set(train_idxs)
-                    & set(test_idxs)
-                ), 'Error: train and test overlap'
+                assert not bool(set(train_idxs) & set(test_idxs)), \
+                    'Error: train and test overlap'
 
                 train = []
                 for pid, idx in enumerate(train_idxs):

torchreid/data/datasets/video/mars.py

@@ -98,7 +98,8 @@ class Mars(VideoDataset):
             if pid == -1:
                 continue # junk images are just ignored
             assert 1 <= camid <= 6
-            if relabel: pid = pid2label[pid]
+            if relabel:
+                pid = pid2label[pid]
             camid -= 1 # index starts from 0
             img_names = names[start_index - 1:end_index]
 

torchreid/data/transforms.py

@@ -4,7 +4,9 @@ import random
 from collections import deque
 import torch
 from PIL import Image
-from torchvision.transforms import *
+from torchvision.transforms import (
+    Resize, Compose, ToTensor, Normalize, ColorJitter, RandomHorizontalFlip
+)
 
 
 class Random2DTranslation(object):
@@ -279,8 +281,13 @@ def build_transforms(
         transform_tr += [RandomHorizontalFlip()]
 
     if 'random_crop' in transforms:
-        print('+ random crop (enlarge to {}x{} and ' \
-              'crop {}x{})'.format(int(round(height*1.125)), int(round(width*1.125)), height, width))
+        print(
+            '+ random crop (enlarge to {}x{} and '
+            'crop {}x{})'.format(
+                int(round(height * 1.125)), int(round(width * 1.125)), height,
+                width
+            )
+        )
         transform_tr += [Random2DTranslation(height, width)]
 
     if 'random_patch' in transforms:

torchreid/engine/engine.py

@@ -428,8 +428,7 @@ class Engine(object):
         if visrank:
             visualize_ranked_results(
                 distmat,
-                self.datamanager.
-                return_query_and_gallery_by_name(dataset_name),
+                self.datamanager.fetch_qg(dataset_name),
                 self.datamanager.data_type,
                 width=self.datamanager.width,
                 height=self.datamanager.height,

torchreid/engine/image/softmax.py

@@ -1,6 +1,4 @@
 from __future__ import division, print_function, absolute_import
-import time
-import datetime
 
 from torchreid import metrics
 from torchreid.losses import CrossEntropyLoss

torchreid/engine/image/triplet.py

@@ -1,6 +1,4 @@
 from __future__ import division, print_function, absolute_import
-import time
-import datetime
 
 from torchreid import metrics
 from torchreid.losses import TripletLoss, CrossEntropyLoss

torchreid/engine/video/triplet.py

@@ -22,11 +22,11 @@ class VideoTripletEngine(ImageTripletEngine, VideoSoftmaxEngine):
             Default is "avg" (average). Choices are ["avg", "max"].
 
     Examples::
-        
+
         import torch
         import torchreid
         # Each batch contains batch_size*seq_len images
-        # Each identity is sampled with num_instances tracklets 
+        # Each identity is sampled with num_instances tracklets
         datamanager = torchreid.data.VideoDataManager(
             root='path/to/reid-data',
             sources='mars',

torchreid/metrics/distance.py

@@ -57,8 +57,9 @@ def euclidean_squared_distance(input1, input2):
         torch.Tensor: distance matrix.
     """
     m, n = input1.size(0), input2.size(0)
-    distmat = torch.pow(input1, 2).sum(dim=1, keepdim=True).expand(m, n) + \
-              torch.pow(input2, 2).sum(dim=1, keepdim=True).expand(n, m).t()
+    mat1 = torch.pow(input1, 2).sum(dim=1, keepdim=True).expand(m, n)
+    mat2 = torch.pow(input2, 2).sum(dim=1, keepdim=True).expand(n, m).t()
+    distmat = mat1 + mat2
     distmat.addmm_(1, -2, input1, input2.t())
     return distmat
 

torchreid/models/hacnn.py

@@ -300,7 +300,8 @@ class HACNN(nn.Module):
         theta = torch.zeros(theta_i.size(0), 2, 3)
         theta[:, :, :2] = scale_factors
         theta[:, :, -1] = theta_i
-        if self.use_gpu: theta = theta.cuda()
+        if self.use_gpu:
+            theta = theta.cuda()
         return theta
 
     def forward(self, x):

torchreid/models/inceptionresnetv2.py

@@ -249,9 +249,9 @@ class Block8(nn.Module):
         return out
 
 
-##################### Model Definition #########################
-
-
+# ----------------
+# Model Definition
+# ----------------
 class InceptionResNetV2(nn.Module):
     """Inception-ResNet-V2.
 

torchreid/models/mlfn.py

@@ -260,7 +260,7 @@ def init_pretrained_weights(model, model_url):
 def mlfn(num_classes, loss='softmax', pretrained=True, **kwargs):
     model = MLFN(num_classes, loss, **kwargs)
     if pretrained:
-        #init_pretrained_weights(model, model_urls['imagenet'])
+        # init_pretrained_weights(model, model_urls['imagenet'])
         import warnings
         warnings.warn(
             'The imagenet pretrained weights need to be manually downloaded from {}'

torchreid/models/mobilenetv2.py

@@ -246,7 +246,7 @@ def mobilenetv2_x1_0(num_classes, loss, pretrained=True, **kwargs):
         **kwargs
     )
     if pretrained:
-        #init_pretrained_weights(model, model_urls['mobilenetv2_x1_0'])
+        # init_pretrained_weights(model, model_urls['mobilenetv2_x1_0'])
         import warnings
         warnings.warn(
             'The imagenet pretrained weights need to be manually downloaded from {}'
@@ -265,7 +265,7 @@ def mobilenetv2_x1_4(num_classes, loss, pretrained=True, **kwargs):
         **kwargs
     )
     if pretrained:
-        #init_pretrained_weights(model, model_urls['mobilenetv2_x1_4'])
+        # init_pretrained_weights(model, model_urls['mobilenetv2_x1_4'])
         import warnings
         warnings.warn(
             'The imagenet pretrained weights need to be manually downloaded from {}'

torchreid/models/nasnet.py

@@ -26,7 +26,7 @@ Code imported from https://github.com/Cadene/pretrained-models.pytorch
 pretrained_settings = {
     'nasnetamobile': {
         'imagenet': {
-            #'url': 'https://github.com/veronikayurchuk/pretrained-models.pytorch/releases/download/v1.0/nasnetmobile-7e03cead.pth.tar',
+            # 'url': 'https://github.com/veronikayurchuk/pretrained-models.pytorch/releases/download/v1.0/nasnetmobile-7e03cead.pth.tar',
             'url':
             'http://data.lip6.fr/cadene/pretrainedmodels/nasnetamobile-7e03cead.pth',
             'input_space': 'RGB',

torchreid/models/shufflenet.py

@@ -45,7 +45,8 @@ class Bottleneck(nn.Module):
         assert stride in [1, 2], 'Warning: stride must be either 1 or 2'
         self.stride = stride
         mid_channels = out_channels // 4
-        if stride == 2: out_channels -= in_channels
+        if stride == 2:
+            out_channels -= in_channels
         # group conv is not applied to first conv1x1 at stage 2
         num_groups_conv1x1 = num_groups if group_conv1x1 else 1
         self.conv1 = nn.Conv2d(
@@ -71,7 +72,8 @@ class Bottleneck(nn.Module):
             mid_channels, out_channels, 1, groups=num_groups, bias=False
         )
         self.bn3 = nn.BatchNorm2d(out_channels)
-        if stride == 2: self.shortcut = nn.AvgPool2d(3, stride=2, padding=1)
+        if stride == 2:
+            self.shortcut = nn.AvgPool2d(3, stride=2, padding=1)
 
     def forward(self, x):
         out = F.relu(self.bn1(self.conv1(x)))
@@ -187,7 +189,7 @@ def init_pretrained_weights(model, model_url):
 def shufflenet(num_classes, loss='softmax', pretrained=True, **kwargs):
     model = ShuffleNet(num_classes, loss, **kwargs)
     if pretrained:
-        #init_pretrained_weights(model, model_urls['imagenet'])
+        # init_pretrained_weights(model, model_urls['imagenet'])
         import warnings
         warnings.warn(
             'The imagenet pretrained weights need to be manually downloaded from {}'

torchreid/models/squeezenet.py

@@ -5,7 +5,6 @@ from __future__ import division, absolute_import
 import torch
 import torch.nn as nn
 import torch.utils.model_zoo as model_zoo
-from torch.utils import model_zoo as model_zoo
 
 __all__ = ['squeezenet1_0', 'squeezenet1_1', 'squeezenet1_0_fc512']
 

torchreid/utils/model_complexity.py

@@ -32,7 +32,7 @@ Convolution
 def hook_convNd(m, x, y):
     k = torch.prod(torch.Tensor(m.kernel_size)).item()
     cin = m.in_channels
-    flops_per_ele = k * cin #+ (k*cin-1)
+    flops_per_ele = k * cin # + (k*cin-1)
     if m.bias is not None:
         flops_per_ele += 1
     flops = flops_per_ele * y.numel() / m.groups
@@ -200,7 +200,7 @@ Linear
 
 
 def hook_linear(m, x, y):
-    flops_per_ele = m.in_features #+ (m.in_features-1)
+    flops_per_ele = m.in_features # + (m.in_features-1)
     if m.bias is not None:
         flops_per_ele += 1
     flops = flops_per_ele * y.numel()

torchreid/utils/rerank.py

@@ -5,8 +5,8 @@ Source: https://github.com/zhunzhong07/person-re-ranking
 
 Created on Mon Jun 26 14:46:56 2017
 @author: luohao
-Modified by Houjing Huang, 2017-12-22. 
-- This version accepts distance matrix instead of raw features. 
+Modified by Houjing Huang, 2017-12-22.
+- This version accepts distance matrix instead of raw features.
 - The difference of `/` division between python 2 and 3 is handled.
 - numpy.float16 is replaced by numpy.float32 for numerical precision.
 

torchreid/utils/torchtools.py

@@ -70,7 +70,7 @@ def load_checkpoint(fpath):
     Returns:
         dict
 
-    Examples::  
+    Examples::
         >>> from torchreid.utils import load_checkpoint
         >>> fpath = 'log/my_model/model.pth.tar-10'
         >>> checkpoint = load_checkpoint(fpath)