Image-to-image transformation in Prompt Engineering / GenAI - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is image-to-image transformation in AI?

Image-to-image transformation is a process where a model takes an input image and changes it into another image, often with a different style or content, like turning a sketch into a photo.

intermediate

Name a common model architecture used for image-to-image transformation.

A common model is the Generative Adversarial Network (GAN), especially variants like Pix2Pix, which learn to convert images from one type to another.

intermediate

What role does the 'generator' play in a GAN for image-to-image tasks?

The generator creates new images from input images trying to fool the discriminator into thinking they are real transformed images.

beginner

Why is paired data important in supervised image-to-image transformation?

Paired data means each input image has a matching target image, helping the model learn exactly how to transform one into the other.

intermediate

What metric can be used to measure the quality of image-to-image transformation?

Metrics like Mean Squared Error (MSE) or Structural Similarity Index (SSIM) compare the transformed image to the target to check quality.

Which model is commonly used for image-to-image transformation?

A Linear Regression

B Decision Tree

C K-Nearest Neighbors

D Generative Adversarial Network (GAN)

GANs are designed to generate new images and are widely used for image-to-image tasks.

What does the discriminator do in a GAN?

A Classifies images as real or fake

B Measures image brightness

C Preprocesses input images

D Creates new images

The discriminator learns to tell apart real images from those created by the generator.

Why is paired data useful in image-to-image transformation?

A It speeds up training by ignoring labels

B It reduces the size of the dataset

C It helps the model learn exact input-output mappings

D It increases randomness in outputs

Paired data provides clear examples of how input images should be transformed.

Which metric compares structural similarity between images?

A Mean Squared Error (MSE)

B Structural Similarity Index (SSIM)

C Accuracy

D Precision

SSIM measures how similar two images are in structure and appearance.

Image-to-image transformation can be used to:

A Turn sketches into photos

B Translate text to speech

C Classify emails as spam

D Predict stock prices

Image-to-image transformation changes one image type into another, like sketches to photos.

Explain how a GAN works for image-to-image transformation.

Think of a game where one player makes images and the other tries to spot fakes.

Describe why paired data is important in supervised image-to-image tasks.

Imagine having before and after photos to learn from.