merge_datasets.py

import os

# Define the paths for both datasets
db_1_path = '/home/ahanaf/Hackathon/Activity_OD/db_1'
db_2_path = '/home/ahanaf/Hackathon/Activity_OD/db_2'

## Paths to the train.txt and Validation.txt files in both datasets
db_1_train_file = os.path.join(db_1_path, 'data', 'images', 'train.txt')
db_2_train_file = os.path.join(db_2_path, 'data', 'images', 'train.txt')
db_1_val_file = os.path.join(db_1_path, 'data', 'images', 'Validation.txt')
db_2_val_file = os.path.join(db_2_path, 'data', 'images', 'Validation.txt')

# Define the merged output file paths
merged_train_file = os.path.join(db_1_path, 'data', 'images', 'merged_train.txt')
merged_val_file = os.path.join(db_1_path, 'data', 'images', 'merged_Validation.txt')

# Function to merge txt files (train/val) and use absolute paths
def merge_txt_files(file1, file2, output_file, base_dir):
    with open(file1, 'r') as f1, open(file2, 'r') as f2, open(output_file, 'w') as out:
        # Read the contents of both files
        f1_lines = f1.readlines()
        f2_lines = f2.readlines()

        # Convert relative paths to absolute paths based on the dataset paths
        f1_lines = [f'/home/ahanaf/Hackathon/Activity_OD/db_1/{line}' for line in f1_lines]
        f2_lines = [f'/home/ahanaf/Hackathon/Activity_OD/db_2/{line}' for line in f2_lines]
        # print(f1_lines[0])
        # exit()

        # Combine the absolute paths and write to the output file
        out.writelines(f1_lines)
        out.writelines(f2_lines)

# Merge the train and validation txt files
merge_txt_files(db_1_train_file, db_2_train_file, merged_train_file, 'images')
merge_txt_files(db_1_val_file, db_2_val_file, merged_val_file, 'images')

print(f"✅ Merged train.txt into {merged_train_file}")
print(f"✅ Merged Validation.txt into {merged_val_file}")

# Now merge the data.yaml files
merged_data_yaml = os.path.join(db_1_path, 'data', 'merged_data.yaml')

with open(merged_data_yaml, 'w') as f:
    # Write the merged data.yaml with appropriate paths
    f.write(f'path: /home/ahanaf/db_1\n')
    f.write(f'train: data/images/merged_train.txt\n')
    f.write(f'val: data/images/merged_Validation.txt\n')
    f.write(f'names:\n')
    f.write(f'  0: AWAY_FROM_DESK\n')
    f.write(f'  1: WORKING\n')
    f.write(f'  2: IDLE\n')

print(f"✅ Merged data.yaml saved at {merged_data_yaml}")