eval.py

import torch
from torchvision import transforms
from vgg import VGG
from datasets import FER2013
from utils import eval, detail_eval
import argparse

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

ap = argparse.ArgumentParser()
ap.add_argument("--dataset_root", default= "fer2013/fer2013.csv", help= "Path to the data folder")
ap.add_argument("--bs", default= 8, type = int,  help = "Batch size for evaluating")
ap.add_argument("--num_workers", default= 4, type= int, help= "Number of workers")
ap.add_argument("--trained_model", default= "model_state.pth.tar", type= str, 
                help = "Trained state_dict file path to open")
ap.add_argument("--model_name", default= "VGG19",type= str, help = "name model")
args= ap.parse_args()

data_path = args.dataset_root
batch_size= args.bs
model_path = args.trained_model
classes = ('Angry', 'Disgust', 'Fear', 'Happy', 'Sad', 'Surprise', 'Neutral')
crop_size= 44
transform_test = transforms.Compose([
        transforms.TenCrop(crop_size),
        transforms.Lambda(lambda crops: torch.stack([transforms.ToTensor()(crop) for crop in crops]))
        ])

trained_model = torch.load(model_path)
print("Load weight model with {} epoch".format(trained_model["epoch"]))

model = VGG(args.model_name)
model.load_state_dict(trained_model["model_weights"])
model.to(device)
model.eval()
publictest_dataset = FER2013(data_path, split= "PUBLIC_TEST", transform= transform_test)
publictest_dataloader = torch.utils.data.DataLoader(publictest_dataset, batch_size= batch_size, 
                                                num_workers= 4)

private_data = FER2013(data_path, split= "PRIVATE_TEST", transform= transform_test)
private_dataloader = torch.utils.data.DataLoader(private_data, batch_size= batch_size, 
                                                num_workers= 4)

print("Evaluation validation (public test) dataset...")
eval(model, publictest_dataloader)
detail_eval(model, publictest_dataloader)
print("-"*10)

print("Evaluation public private dataset...")
eval(model, private_dataloader)
detail_eval(model, private_dataloader)
print("-"*10)