0
0
Computer Visionml~5 mins

Image augmentation transforms in Computer Vision - Cheat Sheet & Quick Revision

Choose your learning style9 modes available
Recall & Review
beginner
What is the purpose of image augmentation in machine learning?
Image augmentation creates new, varied images from existing ones to help models learn better and avoid overfitting by seeing more diverse examples.
Click to reveal answer
beginner
Name three common image augmentation transforms.
Common transforms include rotation (turning the image), flipping (mirroring), and scaling (resizing).
Click to reveal answer
intermediate
How does random rotation help a model learn?
Random rotation shows the model the same object from different angles, helping it recognize objects regardless of orientation.
Click to reveal answer
beginner
What is the difference between horizontal flip and vertical flip?
Horizontal flip mirrors the image left to right, while vertical flip mirrors it top to bottom.
Click to reveal answer
intermediate
Why should augmentation transforms be applied carefully?
Because some transforms can change the meaning of the image (like flipping text), so they must keep the image realistic for the task.
Click to reveal answer
Which of these is NOT a typical image augmentation transform?
ASorting pixels
BFlipping
CAdding noise
DRotation
What does scaling an image do?
AChanges the image color
BChanges the image size
CFlips the image horizontally
DRotates the image
Why use random brightness adjustment in augmentation?
ATo make the image blurry
BTo crop the image
CTo simulate different lighting conditions
DTo flip the image
Which transform would help a model recognize objects regardless of orientation?
ACropping
BNoise addition
CColor inversion
DRotation
What is a risk of applying vertical flip to images with text?
AText becomes unreadable or meaningless
BImage size changes
CColors invert
DModel accuracy improves
Explain why image augmentation is important and list at least three common transforms.
Think about how augmentation helps models see more varied data.
You got /4 concepts.
    Describe how random rotation and flipping help improve model robustness.
    Consider how these transforms simulate real-world variations.
    You got /3 concepts.