Data annotation
I searched for images and carefully identified the main entities present in each image. This involved observing details such as objects, people, actions, and settings, and clearly defining what each entity represented. I then looked for similar entities or moments across different images, ensuring they matched the required criteria and context.