import torch from dataset import YelpDataSet import configure as cf import torchvision.transforms as transforms import matplotlib.pyplot as plt import dataLoader as load if __name__ == "__main__": train, val = load.get_train_valid_loader(cf.photo_url, 50, 32, 'food') print train.__len__() print val.__len__()
import torchvision.models as models from PIL import Image import dataset import dataLoader import configure as cf import plot_utils as utils import train_function as train import resnet as modified_resnet imgTransform = transforms.Compose([transforms.Scale(256), transforms.CenterCrop(224), transforms.ToTensor(), transforms.Normalize((0.4914, 0.4822, 0.4465), (0.2023, 0.1994, 0.2010))]) trainLoader, valLoader = dataLoader.get_train_valid_loader(cf.photo_url,1,32,'all',imgTransform,0.1,-1) # define classification network classification = modified_resnet.resnet50(pretrained = True) classification.fc = nn.Linear(512*4, 5) classification.load_state_dict(torch.load('./res_clas_do7')) #define regression_food regress_food = modified_resnet.resnet50(pretrained = True) regress_food.fc = nn.Linear(512*4, 1) regress_food.load_state_dict(torch.load('./test_FOOD_ResNet_Final')) #define regression_drink regress_drink = models.resnet50(pretrained = False) regress_drink.fc = nn.Linear(512*4, 1) regress_drink.load_state_dict(torch.load('./test_DRINK_ResNet_Final')) #define regression_inside
import dataLoader import configure as cf import train_function as train import plot_utils as utils # define transform function, define trainset and valset # VGG-16 requires the input size of 224*224*3 imgTransform = transforms.Compose( [transforms.Scale(224), transforms.CenterCrop(224), transforms.ToTensor()]) trainLoader, valLoader = \ dataLoader.get_train_valid_loader(cf.photo_url, 50, 32, 'food', imgTransform, 0.1, -1) # define learningRate learningRate = 5 * 1e-4 # Definition of our network. network = models.vgg16(pretrained=True) network.classifier = nn.Sequential( nn.Linear(512 * 7 * 7, 4096), nn.ReLU(True), nn.Dropout(), nn.Linear(4096, 4096), nn.ReLU(True), nn.Dropout(), nn.Linear(4096, 1), )