Changes

README.md

@@ -1 +1,53 @@
-# myPCL
+# myPCL
+
+## Training:
+#### 参数解析
+| 简写参数 |      全称参数       | 描述                                     |
+|:----:|:---------------:|:---------------------------------------|
+|  -a  |     --arch      | 指定主干网络类型，如：resnet-18，resnet-50         |
+|  -j  |    --workers    | 指定线程数，默认为4                             |
+|      |    --epochs     | 总训练循环次数，默认为200                         |
+|      | --warmup-epoch  | 有监督epoch次数，默认为100                      |
+|      |    --exp-dir    | 输出路径，默认为experiment                     |
+|  -b  |  --batch-size   | 一批的数量，默认为8，必须为标签数的倍数                   |
+| -lr  | --learning-rate | 学习率，默认为0.03                            |
+|      |      --cos      | 使用cosine学习率                            |
+|      |   --schedule    | 指定学习率下降的epoch，默认为[120,160]，只在cos未指定时生效 |
+|      |   --momentum    | 优化器的动量餐宿，默认为0.9                        |
+| --wd | --weight-decay  | SSPCL模型的权重衰减，默认为1e-4                   |
+|      |    --low-dim    | 输出维度，默认为128                            |
+|      |  --num-cluster  | 聚类个数，默认为'20,25,30'                     |
+|      |     --pcl-r     | 负例对，需要小于聚类个数，默认为16                     |
+|      |    --moco-m     | SSPCL中ME更新参数使用的动量，默认为0.999             |
+|      |      --mlp      | 设置即为使用mlp，无参数，参考PCL模型                  |
+|      |  --temperature  | softmax层温度参数，默认为0.2                    |
+|  -p  |  --print-freq   | 显示频率，默认为每10个数据                         |
+|      |   --save-freq   | 保存模型的频率，默认为每10个epoch                   |
+|      |  --world-size   | 总训练程序数量，默认为1                           |
+|      |     --rank      | 此训练程序编号，默认编号0                          |
+|      |   --dist-url    | 多程序训练连接地址，此参数参照pytorch分布式训练解释          |
+|      | --dist-backend  | 默认为nccl                                |
+|      |      --gpu      | 用于训练的gpu编号                             |
+|      |     --seed      | 随机数种子，默认为自动生成                          |
+|      |    --resume     | 需要载入的模型位置                              |
+|      |  --start-epoch  | 训练起始的epoch，与resume配合使用                 |
+|      |                 |                                        |
+
+#### 用例
+<pre>
+python main.py -a resnet18 --lr 0.03 --batch-size 8 --workers 4 --temperature 0.2 --mlp --aug-plus --cos --dist-url "tcp://localhost:10001" --world-size 1 --rank 0 --warmup-epoch 100 --epochs 100 --exp-dir exp images
+</pre>
+
+## Testing:
+#### 参数解析
+**如训练时修改了以上默认的参数，在测试时也需要指定**
+以下是必须要设置的参数
+
+| 简写参数 |     全称参数     | 描述        |
+|:----:|:------------:|:----------|
+|      | --pretrained | 需要载入模型的路径 |
+
+#### 用例
+<pre>
+python test_svm.py --pretrained exp/checkpoint_0199.pth.tar
+</pre>

main.py

@@ -15,8 +15,6 @@ import torch.optim
 import torch.multiprocessing as mp
 import torch.utils.data
 import torch.utils.data.distributed
-import torchvision.transforms as transforms
-import torchvision.datasets as datasets
 import torchvision.models as models
 
 from scripts.parser import parser
@@ -24,11 +22,13 @@ from scripts.meter import AverageMeter, ProgressMeter
 import scripts.augmentation as aug
 import scripts.momentum as momentum
 import scripts.clustering as clustering
+import scripts.loss as loss_script
+from scripts.data_process import data_process
 import pcl.builder
 import pcl.loader
 
 
-def main_loader():
+def worker_loader():
     args = parser().parse_args()
 
     if args.seed is not None:
@@ -61,14 +61,15 @@ def main_loader():
         args.world_size = ngpus_per_node * args.world_size
         # Use torch.multiprocessing.spawn to launch distributed processes: the
         # main_worker process function
-        mp.spawn(main_worker, nprocs=ngpus_per_node, args=(ngpus_per_node, args))
+        mp.spawn(worker, nprocs=ngpus_per_node, args=(ngpus_per_node, args))
     else:
         # Simply call main_worker function
-        main_worker(args.gpu, ngpus_per_node, args)
+        worker(args.gpu, ngpus_per_node, args)
 
 
-def main_worker(gpu, ngpus_per_node, args):
+def worker(gpu, ngpus_per_node, args):
     args.gpu = gpu
+    args.multiprocessing_distributed = True
 
     if args.gpu is not None:
         print("Use GPU: {} for training".format(args.gpu))
@@ -157,13 +158,6 @@ def main_worker(gpu, ngpus_per_node, args):
     train_dir = os.path.join(args.data, 'train')
     eval_dir = os.path.join(args.data, 'train')
 
-    if args.aug_plus:
-        # MoCo v2's aug: similar to SimCLR https://arxiv.org/abs/2002.05709
-        augmentation = aug.moco_v2()
-    else:
-        # MoCo v1's aug: same as InstDisc https://arxiv.org/abs/1805.01978
-        augmentation = aug.moco_v1()
-
     # center-crop augmentation
     eval_augmentation = aug.moco_eval()
 
@@ -248,12 +242,12 @@ def main_worker(gpu, ngpus_per_node, args):
 
         if (epoch + 1) % args.save_freq == 0 and (not args.multiprocessing_distributed or (args.multiprocessing_distributed
                                                                                and args.rank % ngpus_per_node == 0)):
-            save_checkpoint({
+            torch.save({
                 'epoch': epoch + 1,
                 'arch': args.arch,
                 'state_dict': model.state_dict(),
                 'optimizer': optimizer.state_dict(),
-            }, is_best=False, filename='{}/checkpoint_{:04d}.pth.tar'.format(args.exp_dir, epoch))
+            }, '{}/checkpoint_{:04d}.pth.tar'.format(args.exp_dir, epoch))
 
 
 def train(train_loader, model, criterion, optimizer, epoch, args, cluster_result=None):
@@ -276,46 +270,18 @@ def train(train_loader, model, criterion, optimizer, epoch, args, cluster_result
         # measure data loading time
         data_time.update(time.time() - end)
 
-        im_q = []
+        im_q, im_k = data_process(cluster_result, images, args.gpu)
-        im_k = []
-        if cluster_result is None:
-            class_number = len(images)
-            class_len = len(images[0])
-            for _i in range(0, class_len, 2):
-                for c in range(class_number):
-                    im_q.append(images[c][_i])
-                    im_k.append(images[c][_i+1])
-            im_q = torch.stack(im_q)
-            im_k = torch.stack(im_k)
-        else:
-            im_q = images[0]
-            im_k = images[1]
-
-        if args.gpu is not None:
-            im_q = im_q.cuda(args.gpu, non_blocking=True)
-            im_k = im_k.cuda(args.gpu, non_blocking=True)
 
         # compute output
         output, target, output_proto, target_proto = model(im_q=im_q, im_k=im_k,
                                                            cluster_result=cluster_result, index=index)
 
-        # InfoNCE loss
+        loss = loss_script.proto_with_quality(output, target, output_proto, target_proto, criterion, acc_proto, images,
-        loss = criterion(output, target)
+                                              args.num_cluster)
 
-        # ProtoNCE loss
-        if output_proto is not None:
-            loss_proto = 0
-            for proto_out, proto_target in zip(output_proto, target_proto):
-                loss_proto += criterion(proto_out, proto_target)
-                accp = accuracy(proto_out, proto_target)[0]
-                acc_proto.update(accp[0], images[0].size(0))
-
-            # average loss across all sets of prototypes
-            loss_proto /= len(args.num_cluster)
-            loss += loss_proto
 
         losses.update(loss.item(), images[0].size(0))
-        acc = accuracy(output, target)[0]
+        acc = loss_script.accuracy(output, target)[0]
         acc_inst.update(acc[0], images[0].size(0))
 
         # compute gradient and do SGD step
@@ -343,29 +309,5 @@ def adjust_learning_rate(optimizer, epoch, args):
         param_group['lr'] = lr
 
 
-def accuracy(output, target, topk=(1,)):
-    """Computes the accuracy over the k top predictions for the specified values of k"""
-    with torch.no_grad():
-        maxk = max(topk)
-        batch_size = target.size(0)
-
-        _, pred = output.topk(maxk, 1, True, True)
-        pred = pred.t()
-        correct = pred.eq(target.view(1, -1).expand_as(pred))
-
-        res = []
-        for k in topk:
-            correct_k = correct[:k].view(-1).float().sum(0, keepdim=True)
-            res.append(correct_k.mul_(100.0 / batch_size))
-        return res
-
-
-def save_checkpoint(state, is_best, filename='checkpoint.pth.tar'):
-    torch.save(state, filename)
-    print("Model saved as:"+filename)
-    if is_best:
-        shutil.copyfile(filename, 'model_best.pth.tar')
-
-
 if __name__ == '__main__':
-    main_loader()
+    worker_loader()

pcl/loader.py

@@ -35,6 +35,13 @@ class PreImager(tud.Dataset):
     def __len__(self):
         return self.length
 
+class TenserImager(datasets.ImageFolder):
+    def __getitem__(self, index):
+        path, target = self.samples[index]
+        sample = self.loader(path)
+        sample = self.transform(sample)
+        return sample, target
+
 
 class TwoCropsTransform:
     """Take two random crops of one image as the query and key."""

scripts/data_process.py

@@ -0,0 +1,22 @@
+import torch
+def data_process(cluster_result, images, gpu):
+    im_q = []
+    im_k = []
+    if cluster_result is None:
+        class_number = len(images)
+        class_len = len(images[0])
+        for _i in range(0, class_len, 2):
+            for c in range(class_number):
+                im_q.append(images[c][_i])
+                im_k.append(images[c][_i + 1])
+        im_q = torch.stack(im_q)
+        im_k = torch.stack(im_k)
+    else:
+        im_q = images[0]
+        im_k = images[1]
+
+    if gpu is not None:
+        im_q = im_q.cuda(gpu, non_blocking=True)
+        im_k = im_k.cuda(gpu, non_blocking=True)
+
+    return im_q, im_k

scripts/loss.py

@@ -0,0 +1,42 @@
+import torch
+import numpy as np
+import math
+def proto_with_quality(output, target, output_proto, target_proto, criterion, acc_proto, images, num_cluster):
+    # InfoNCE loss
+    loss = criterion(output, target)
+
+    # ProtoNCE loss
+    if output_proto is not None:
+        loss_proto = 0
+        for proto_out, proto_target in zip(output_proto, target_proto):
+            loss_proto += criterion(proto_out, proto_target)
+            accp = accuracy(proto_out, proto_target)[0]
+            acc_proto.update(accp[0], images[0].size(0))
+
+        # average loss across all sets of prototypes
+        loss_proto /= len(num_cluster)
+        loss += loss_proto
+
+        # Quality loss
+        mse = np.mean((images[1]/255.0-images[2]/255.0)**2)
+        psnr = 20 * math.log10(1 / math.sqrt(mse))
+
+
+    return loss
+
+
+def accuracy(output, target, topk=(1,)):
+    """Computes the accuracy over the k top predictions for the specified values of k"""
+    with torch.no_grad():
+        maxk = max(topk)
+        batch_size = target.size(0)
+
+        _, pred = output.topk(maxk, 1, True, True)
+        pred = pred.t()
+        correct = pred.eq(target.view(1, -1).expand_as(pred))
+
+        res = []
+        for k in topk:
+            correct_k = correct[:k].view(-1).float().sum(0, keepdim=True)
+            res.append(correct_k.mul_(100.0 / batch_size))
+        return res

scripts/parser.py

@@ -14,35 +14,52 @@ def parser():
                          help='model architecture: ' +
                               ' | '.join(model_names) +
                               ' (default: resnet50)')
-    _parser.add_argument('-j', '--workers', default=8, type=int, metavar='N',
+    _parser.add_argument('-j', '--workers', default=4, type=int, metavar='N',
-                         help='number of data loading workers (default: 8)')
+                         help='number of data loading workers (default: 4)')
     _parser.add_argument('--epochs', default=200, type=int, metavar='N',
                          help='number of total epochs to run')
-    _parser.add_argument('--start-epoch', default=0, type=int, metavar='N',
+    _parser.add_argument('--warmup-epoch', default=100, type=int,
-                         help='manual epoch number (useful on restarts)')
+                         help='number of warm-up epochs to only train with InfoNCE loss')
-    _parser.add_argument('-b', '--batch-size', default=256, type=int,
+    _parser.add_argument('--exp-dir', default='experiment', type=str,
+                         help='experiment directory')
+
+    _parser.add_argument('-b', '--batch-size', default=8, type=int,
                          metavar='N',
-                         help='mini-batch size (default: 256), this is the total '
+                         help='mini-batch size (default: 8), this is the total '
                               'batch size of all GPUs on the current node when '
                               'using Data Parallel or Distributed Data Parallel')
-    _parser.add_argument('--lr', '--learning-rate', default=0.03, type=float,
+    _parser.add_argument('-lr', '--learning-rate', default=0.03, type=float,
                          metavar='LR', help='initial learning rate', dest='lr')
+    _parser.add_argument('--cos', action='store_true',
+                         help='use cosine lr schedule')
     _parser.add_argument('--schedule', default=[120, 160], nargs='*', type=int,
                          help='learning rate schedule (when to drop lr by 10x)')
+
     _parser.add_argument('--momentum', default=0.9, type=float, metavar='M',
                          help='momentum of SGD solver')
     _parser.add_argument('--wd', '--weight-decay', default=1e-4, type=float,
                          metavar='W', help='weight decay (default: 1e-4)',
                          dest='weight_decay')
+    _parser.add_argument('--low-dim', default=128, type=int,
+                         help='feature dimension (default: 128)')
+    _parser.add_argument('--num-cluster', default='20,25,30', type=str,
+                         help='number of clusters')
+    _parser.add_argument('--pcl-r', default=16, type=int,
+                         help='queue size; number of negative pairs; needs to be smaller than num_cluster (default: '
+                              '16)')
+    _parser.add_argument('--moco-m', default=0.999, type=float,
+                         help='moco momentum of updating key encoder (default: 0.999)')
+    _parser.add_argument('--temperature', default=0.2, type=float,
+                         help='softmax temperature')
+
     _parser.add_argument('-p', '--print-freq', default=10, type=int,
                          metavar='N', help='print frequency (default: 10)')
     _parser.add_argument('--save-freq', default=10, type=int,
                          metavar='N', help='save frequency (default: 10)')
-    _parser.add_argument('--resume', default='', type=str, metavar='PATH',
+
-                         help='path to latest checkpoint (default: none)')
+    _parser.add_argument('--world-size', default=1, type=int,
-    _parser.add_argument('--world-size', default=-1, type=int,
                          help='number of nodes for distributed training')
-    _parser.add_argument('--rank', default=-1, type=int,
+    _parser.add_argument('--rank', default=0, type=int,
                          help='node rank for distributed training')
     _parser.add_argument('--dist-url', default='tcp://172.0.0.1:23456', type=str,
                          help='url used to set up distributed training')
@@ -58,33 +75,16 @@ def parser():
                               'fastest way to use PyTorch for either single node or '
                               'multi node data parallel training')
 
-    _parser.add_argument('--low-dim', default=128, type=int,
+    _parser.add_argument('--start-epoch', default=0, type=int, metavar='N',
-                         help='feature dimension (default: 128)')
+                         help='manual epoch number (useful on restarts)')
-    _parser.add_argument('--pcl-r', default=16, type=int,
+    _parser.add_argument('--resume', default='', type=str, metavar='PATH',
-                         help='queue size; number of negative pairs; needs to be smaller than num_cluster (default: '
+                         help='path to latest checkpoint (default: none)')
-                              '16384)')
-    _parser.add_argument('--moco-m', default=0.999, type=float,
-                         help='moco momentum of updating key encoder (default: 0.999)')
-    _parser.add_argument('--temperature', default=0.2, type=float,
-                         help='softmax temperature')
 
     _parser.add_argument('--mlp', action='store_true',
                          help='use mlp head')
-    _parser.add_argument('--aug-plus', action='store_true',
-                         help='use moco-v2/SimCLR data augmentation')
-    _parser.add_argument('--cos', action='store_true',
-                         help='use cosine lr schedule')
-
-    _parser.add_argument('--num-cluster', default='20,25,30', type=str,
-                         help='number of clusters')
-    _parser.add_argument('--warmup-epoch', default=20, type=int,
-                         help='number of warm-up epochs to only train with InfoNCE loss')
-    _parser.add_argument('--exp-dir', default='experiment', type=str,
-                         help='experiment directory')
 
     _parser.add_argument('--cost', type=str, default='0.5')
-
+    _parser.add_argument('--n-run', type=int, default=1)
-    _parser.add_argument('--num-class', type=int, default=20)
     _parser.add_argument('--pretrained', default='', type=str,
                         help='path to pretrained checkpoint')
 

test_svm.py

@@ -1,25 +1,27 @@
 from __future__ import print_function
 
 import os
-import sys
-import time
 import torch
-import torch.optim as optim
+from copy import deepcopy
-import torch.backends.cudnn as cudnn
-import torch.nn.functional as F
-import argparse
-import random
 import numpy as np
 
 from torchvision import transforms, datasets
 import torchvision.models as models
 
-from sklearn.svm import LinearSVC
+from sklearn.svm import LinearSVC, SVC
+from sklearn.multiclass import OneVsRestClassifier
+from sklearn.metrics import confusion_matrix, roc_curve, roc_auc_score, precision_score, f1_score
+
+from matplotlib import pyplot as plt
 
 from scripts.parser import parser
 import scripts.augmentation as aug
 import pcl.loader
 
+import matplotlib.font_manager
+# 通过字体文件添加字体
+matplotlib.font_manager.fontManager.addfont('simsun.ttc')
+
 def calculate_ap(rec, prec):
     """
     Computes the AP under the precision recall curve.
@@ -76,9 +78,9 @@ def get_precision_recall(targets, preds):
 def main():
     args = parser().parse_args()
 
-    if not args.seed is None:
+    # if not args.seed is None:
-        random.seed(args.seed)
+    #     random.seed(args.seed)
-        np.random.seed(args.seed)
+    #     np.random.seed(args.seed)
 
     mean = [0.485, 0.456, 0.406]
     std = [0.229, 0.224, 0.225]
@@ -93,10 +95,10 @@ def main():
     eval_augmentation = aug.moco_eval()
 
     pre_train_dir = os.path.join(args.data, 'pre_train')
-    eval_dir = os.path.join(args.data, 'eval')
+    eval_dir = os.path.join(args.data, 'test')
 
-    train_dataset = pcl.loader.PreImager(pre_train_dir, eval_augmentation)
+    train_dataset = pcl.loader.TenserImager(pre_train_dir, eval_augmentation)
-    val_dataset = pcl.loader.ImageFolderInstance(
+    val_dataset = pcl.loader.TenserImager(
         eval_dir,
         eval_augmentation)
 
@@ -162,7 +164,8 @@ def main():
                 train_loader = torch.utils.data.DataLoader(
                     train_dataset, batch_size=args.batch_size, shuffle=False,
                     num_workers=args.workers, pin_memory=True)
-
+                classes = len(train_dataset.classes)
+                classes_name = train_dataset.classes
                 train_feats = []
                 train_labels = []
                 print('==> calculate train features')
@@ -181,27 +184,68 @@ def main():
                 train_feats = train_feats / (train_feats_norm + 1e-5)[:, np.newaxis]
 
                 print('==> training SVM Classifier')
-                cls_ap = np.zeros((args.num_class, 1))
+                #test_labels[test_labels == 0] = -1
-                test_labels[test_labels == 0] = -1
+                #train_labels[train_labels == 0] = -1
-                train_labels[train_labels == 0] = -1
+                clf = OneVsRestClassifier(LinearSVC(
-                for cls in range(args.num_class):
+                    C=cost, # class_weight={1: 2, -1: 1},
-                    clf = LinearSVC(
+                    intercept_scaling=1.0,
-                        C=cost, class_weight={1: 2, -1: 1}, intercept_scaling=1.0,
+                    penalty='l2', loss='squared_hinge', tol=1e-4,
-                        penalty='l2', loss='squared_hinge', tol=1e-4,
+                    dual=True, max_iter=2000, random_state=0))
-                        dual=True, max_iter=2000, random_state=0)
+                clf.fit(train_feats, train_labels)
-                    clf.fit(train_feats, train_labels[:, cls])
 
-                    prediction = clf.decision_function(test_feats)
+                prediction = clf.decision_function(test_feats)
-                    P, R, score, ap = get_precision_recall(test_labels[:, cls], prediction)
+                predict = clf.predict(test_feats)
-                    cls_ap[cls][0] = ap * 100
-                mean_ap = np.mean(cls_ap, axis=0)
 
-                print('==> Run%d mAP is %.2f: ' % (run, mean_ap))
+
+                plt.figure(1)
+                plt.rcParams['font.sans-serif'] = ['simsun']
+                plt.figure(2)
+                plt.rcParams['font.sans-serif'] = ['simsun']
+
+                list_ap = []
+                list_auc = []
+                for cl in range(classes):
+                    t_labels,t_pre = deepcopy(test_labels),deepcopy(predict)
+                    t_labels[t_labels != cl] = -1
+                    t_labels[t_labels == cl] = 1
+                    clf = LinearSVC()
+                    clf.fit(test_feats, t_labels)
+                    t_pre=clf.predict(test_feats)
+
+                    P, R, score, ap = get_precision_recall(t_labels, t_pre)
+                    fpr, tpr, thres =roc_curve(t_labels, t_pre)
+                    auc = roc_auc_score(t_labels, t_pre)
+                    list_ap.append(ap)
+                    list_auc.append(auc)
+                    plt.figure(1)
+                    plt.plot(R, P)
+                    plt.figure(2)
+                    plt.plot(fpr, tpr)
+
+                plt.figure(1)
+                plt.xlabel('召回率', fontsize=14)
+                plt.ylabel('精准率', fontsize=14)
+                plt.legend(classes_name)
+                plt.savefig("PR.png")
+
+                plt.figure(2)
+                plt.xlabel('假正率', fontsize=14)
+                plt.ylabel('真正率', fontsize=14)
+                plt.legend(classes_name)
+                plt.savefig("ROC.png")
+                print(classes_name)
+                confusion = confusion_matrix(test_labels, predict)
+                print(confusion)
+                mean_ap = np.mean(list_ap) * 100
+                print('==> Run%d\nmAP is %f ' % (run, mean_ap))
+                print("AP: " + str(precision_score(test_labels, predict, average="weighted")))
+                print("mAUC: "+ str(np.mean(list_auc)))
+                print("F1_score: " + str(f1_score(test_labels, predict, average="weighted")))
                 avg_map.append(mean_ap)
 
             avg_map = np.asarray(avg_map)
-            print('Cost:%.2f - Average ap is: %.2f' % (cost, avg_map.mean()))
+            #print('Cost:%.2f - Average ap is: %.2f' % (cost, avg_map.mean()))
-            print('Cost:%.2f - Std is: %.2f' % (cost, avg_map.std()))
+            #print('Cost:%.2f - Std is: %.2f' % (cost, avg_map.std()))
             result_k[i] = avg_map.mean()
         result[k] = result_k.max()
     print(result)