utils/create_dataset.py

from torch.utils.data import random_split
from pathlib import Path
import random
from utils.Dataset import augment_sample,augment_data,balance_data
from statistics import mean
import warnings
import torchvision.transforms as transforms
from utils.utils import *

warnings.filterwarnings("ignore")

warnings.filterwarnings("default")


import os

transform = transforms.Compose([
    transforms.RandomHorizontalFlip(),
    transforms.RandomVerticalFlip(),
    transforms.RandomHorizontalFlip(),
    transforms.RandomVerticalFlip(),
])

def create_data_set(
    root_dir,
    train_test_split_dir,
    label_list,
    n_way,
    n_shot,
    n_query,
    env_no,
     
):
    

    train_filenames = []
    train_labels = []
    valid_data = []
    permutations_str = "_".join(map(str, label_list))
    env_str = "_".join(map(str, env_no))
    with open(train_test_split_dir, 'r') as file:
        lines = file.readlines()
        for line in lines:
            parts = line.strip().split()
            if int(parts[-1]) in env_no or parts[-1] == '0':
                train_filenames.append(os.path.join(root_dir, parts[0].split("/")[1]))
                train_labels.append(int(parts[1]))
        # print(train_filenames)
        # Randomly shuffle the data and labels
        train_data = list(zip(train_filenames, train_labels))
        unique_labels = set(train_labels)
        for label in unique_labels:
            label_indices = [i for i, l in enumerate(train_labels) if l == label]
            random.shuffle(label_indices)
            num_valid_samples = int(len(label_indices) * 0.2)
            valid_indices = label_indices[:num_valid_samples]
            valid_data.extend([(train_data[i][0], train_data[i][1]) for i in valid_indices])

        filtered_train_data = [(filename, label) for filename, label in train_data if label in label_list]
        # print(valid_data)
        filtered_valid_data = [(filename, label) for filename, label in valid_data if label in label_list]
        # print(filtered_valid_data)
        filtered_valid_data_balnced = balance_data(n_shot,n_query,filtered_valid_data)
        # train_data = balance_data(n_shot,n_query,train_data)
        save_balanced_data_to_csv(filtered_train_data, f'train_data_{permutations_str}_env{env_str}.csv')
        print(f"file has stored in train_data_{permutations_str}_env{env_no}.csv")
        # valid_data = balance_data(n_shot,n_query,valid_data)
        save_balanced_data_to_csv(filtered_valid_data_balnced, f'valid_data_{permutations_str}_env{env_str}.csv')
        print(f"file has stored in 'valid_data_{permutations_str}_env{env_no}.csv")


def create_data_set_fs(
    root_dir,
    train_test_split_dir,
    label_list,
    n_way,
    n_shot,
    n_query,     
):
    

    train_filenames = []
    train_labels = []
    valid_data = []
    permutations_str = "_".join(map(str, label_list))
    
    with open(train_test_split_dir, 'r') as file:
        lines = file.readlines()
        for line in lines:
            parts = line.strip().split()
            train_filenames.append(os.path.join(root_dir, parts[0].split("/")[1]))
            train_labels.append(int(parts[1]))
        # print(train_filenames)
        # Randomly shuffle the data and labels
        train_data = list(zip(train_filenames, train_labels))
        unique_labels = set(train_labels)
        for label in unique_labels:
            label_indices = [i for i, l in enumerate(train_labels) if l == label]
            random.shuffle(label_indices)
            num_valid_samples = int(len(label_indices) * 0.2)
            valid_indices = label_indices[:num_valid_samples]
            valid_data.extend([(train_data[i][0], train_data[i][1]) for i in valid_indices])

        filtered_train_data = [(filename, label) for filename, label in train_data if label in label_list]
        # print(valid_data)
        filtered_valid_data = [(filename, label) for filename, label in valid_data if label in label_list]
        # print(filtered_valid_data)
        filtered_valid_data_balnced = balance_data(n_shot,n_query,filtered_valid_data)
        # train_data = balance_data(n_shot,n_query,train_data)
        save_balanced_data_to_csv(filtered_train_data, f'train_data_{permutations_str}.csv')
        print(f"file has stored in train_data_{permutations_str}.csv")
        # valid_data = balance_data(n_shot,n_query,valid_data)
        save_balanced_data_to_csv(filtered_valid_data_balnced, f'valid_data_{permutations_str}.csv')
        print(f"file has stored in 'valid_data_{permutations_str}.csv")