train.py

# Imports here
import torch
import torch.nn as nn
import torch.optim as optim
from torch.optim import lr_scheduler
import numpy as np
import time
import os
import argparse
from torchvision import datasets, models, transforms, utils
from torch.autograd import Variable
import torch.nn.functional as F
import copy
from PIL import Image

#Globals
nThreads = 4
batch_size = 8
use_gpu = torch.cuda.is_available()

def cook_data(args):
    
    data_dir = args.data_dir
    train_dir = data_dir + '/train'
    valid_dir = data_dir + '/valid'
    test_dir = data_dir + '/test'

    #Define your transforms for the training, validation, and testing sets
    train_transforms = transforms.Compose([transforms.RandomRotation(30),
                                       transforms.RandomResizedCrop(224),
                                       transforms.RandomHorizontalFlip(),
                                       transforms.ToTensor(),
                                       transforms.Normalize([0.485, 0.456, 0.406], 
                                                            [0.229, 0.224, 0.225])])

    valid_transforms = transforms.Compose([transforms.Resize(256),
                                          transforms.CenterCrop(224),
                                          transforms.ToTensor(),
                                          transforms.Normalize([0.485, 0.456, 0.406], 
                                                               [0.229, 0.224, 0.225])])

    test_transforms = transforms.Compose([transforms.Resize(256),
                                          transforms.CenterCrop(224),
                                          transforms.ToTensor(),
                                          transforms.Normalize([0.485, 0.456, 0.406], 
                                                               [0.229, 0.224, 0.225])])

    # TODO: Load the datasets with ImageFolder
    image_datasets = dict()
    image_datasets['train'] = datasets.ImageFolder(train_dir, transform=train_transforms)
    image_datasets['valid'] = datasets.ImageFolder(valid_dir, transform=valid_transforms)
    image_datasets['test'] = datasets.ImageFolder(test_dir, transform=test_transforms)


    # TODO: Using the image datasets and the trainforms, define the dataloaders

    dataloaders = dict()
    dataloaders['train'] = torch.utils.data.DataLoader(image_datasets['train'], batch_size=batch_size, shuffle=True)
    dataloaders['valid'] = torch.utils.data.DataLoader(image_datasets['valid'], batch_size=batch_size)
    dataloaders['test']  = torch.utils.data.DataLoader(image_datasets['test'], batch_size=batch_size)

    return dataloaders, image_datasets

def train_model(args, model, criterion, optimizer, scheduler, num_epochs=25):
    
    dataloaders, image_datasets = cook_data(args)
    dataset_sizes = {x: len(image_datasets[x]) for x in ['train', 'valid', 'test']}
    since = time.time()

    best_model_wts = copy.deepcopy(model.state_dict())
    best_acc = 0.0

    for epoch in range(num_epochs):
        print('Epoch {}/{}'.format(epoch, num_epochs - 1))
        print('-' * 10)

        # Each epoch has a training and validation phase
        for phase in ['train', 'valid']:
            if phase == 'train':
                scheduler.step()
                model.train(True)  # Set model to training mode
            else:
                model.train(False)  # Set model to evaluate mode

            running_loss = 0.0
            running_corrects = 0

            # Iterate over data.
            for data in dataloaders[phase]:
                # get the inputs
                inputs, labels = data

                # wrap them in Variable
                if use_gpu and args.gpu:
                    inputs = Variable(inputs.cuda())
                    labels = Variable(labels.cuda())
                else:
                    inputs, labels = Variable(inputs), Variable(labels)

                # zero the parameter gradients
                optimizer.zero_grad()

                # forward
                outputs = model(inputs)
                _, preds = torch.max(outputs.data, 1)
                loss = criterion(outputs, labels)

                # backward + optimize only if in training phase
                if phase == 'train':
                    loss.backward()
                    optimizer.step()

                # statistics
                running_loss += loss.data[0] * inputs.size(0)
                running_corrects += torch.sum(preds == labels.data)

            epoch_loss = running_loss / dataset_sizes[phase]
            epoch_acc = running_corrects / dataset_sizes[phase]

            print('{} Loss: {:.4f} Acc: {:.4f}'.format(
                phase, epoch_loss, epoch_acc))

            # deep copy the model
            if phase == 'valid' and epoch_acc > best_acc:
                best_acc = epoch_acc
                best_model_wts = copy.deepcopy(model.state_dict())

        print()

    time_elapsed = time.time() - since
    print('Training complete in {:.0f}m {:.0f}s'.format(
        time_elapsed // 60, time_elapsed % 60))
    print('Best val Acc: {:4f}'.format(best_acc))

    # load best model weights
    model.load_state_dict(best_model_wts)
    return model


def train_model_wrapper(args):

    dataloaders, image_datasets = cook_data(args)

    # 1. Load a pre-trained network
    if args.arch == 'vgg': 
        model = models.vgg16(pretrained=True)
    elif args.arch == 'densenet':
        model = models.densenet121(pretrained=True)

    # Freeze parameters so we don't backprop through them
    for param in model.parameters():
        param.requires_grad = False

    # 2. Define a new, untrained feed-forward network as a classifier, using ReLU activations and dropout

    num_features = model.classifier[0].in_features
    from collections import OrderedDict
    classifier = nn.Sequential(OrderedDict([
                              ('fc1', nn.Linear(num_features, 512)),
                              ('relu', nn.ReLU()),
                              ('drpot', nn.Dropout(p=0.5)),
                              ('hidden', nn.Linear(512, args.hidden_units)),                       
                              ('fc2', nn.Linear(args.hidden_units, 102)),
                              ('output', nn.LogSoftmax(dim=1)),
                              ]))

    # Reserve for final layer: ('output', nn.LogSoftmax(dim=1))
        
    model.classifier = classifier

    # 3. Train the classifier layers using backpropagation using the pre-trained network to get the features
    # 4. Track the loss and accuracy on the validation set to determine the best hyperparameters
        
    if args.gpu:
        if use_gpu:
            model = model.cuda()
            print ("Using GPU: "+ str(use_gpu))
        else:
            print("Using CPU since GPU is not available/configured")


    criterion = nn.CrossEntropyLoss()
    optimizer = optim.Adam(model.classifier.parameters(), lr=args.lr)
    exp_lr_scheduler = lr_scheduler.StepLR(optimizer, step_size=7, gamma=0.1)
    
    model = train_model(args, model, criterion, optimizer, exp_lr_scheduler,num_epochs=args.epochs)

    model.class_to_idx = dataloaders['train'].dataset.class_to_idx
    model.epochs = args.epochs
    checkpoint = {'input_size': [3, 224, 224],
                  'batch_size': dataloaders['train'].batch_size,
                  'output_size': 102,
                  'arch': args.arch,
                  'state_dict': model.state_dict(),
                  'optimizer_dict':optimizer.state_dict(),
                  'class_to_idx': model.class_to_idx,
                  'epoch': model.epochs}
    torch.save(checkpoint, args.saved_model)


def main():
    parser = argparse.ArgumentParser(description='Flower Classifcation trainer')
    parser.add_argument('--gpu', type=bool, default=False, help='Use GPU or not')
    parser.add_argument('--arch', type=str, default='densenet', help='architecture [available: densenet, vgg]', required=True)
    parser.add_argument('--lr', type=float, default=0.001, help='learning rate')
    parser.add_argument('--hidden_units', type=int, default=100, help='hidden units for fc layer')
    parser.add_argument('--epochs', type=int, default=15, help='number of epochs')
    parser.add_argument('--data_dir', type=str, default='flowers', help='dataset directory')
    parser.add_argument('--saved_model' , type=str, default='my_checkpoint_cmd.pth', help='path of your saved model')
    args = parser.parse_args()

    import json
    with open('cat_to_name.json', 'r') as f:
        cat_to_name = json.load(f)

    train_model_wrapper(args)


if __name__ == "__main__":
    main()