Computer Vision Flashcards
6 cards from real Artificial Intelligence practice questions. Tap to flip, then mark Knew It or Still Learning — missed cards come back until you master them.
Read the first 6 Computer Vision flashcards as text
What is semantic segmentation in computer vision?
Answer: Assigning a class label to every pixel in the image without distinguishing between instances
Semantic segmentation classifies every pixel into a category (road, sky, car) but treats all instances of the same class as one region, unlike instance segmentation.
Which architecture introduced the concept of encoder-decoder with skip connections, widely used for image segmentation?
Answer: U-Net
U-Net uses a contracting encoder path and an expansive decoder path with skip connections between corresponding layers, enabling precise pixel-level segmentation.
What is the purpose of mean Average Precision (mAP) in evaluating object detectors?
Answer: Averaging precision scores over all classes and IoU thresholds to give a single detection quality metric
mAP averages the area under the precision-recall curve across all object categories, providing a comprehensive summary of detector performance.
What is 'image augmentation' with random cropping designed to help the model learn?
Answer: To make predictions invariant to the object's position and scale within the image
Random cropping exposes the model to objects appearing at different positions and scales, building spatial invariance into the learned features.
In a generative model like a Variational Autoencoder (VAE) for images, what does the latent space represent?
Answer: A compressed, continuous lower-dimensional representation that captures the essential factors of variation in the data
The VAE's latent space encodes images as probability distributions over a compact representation, from which new samples can be decoded to generate novel images.
What is 'image recognition' and how does it differ from 'object detection'?
Answer: Image recognition assigns a label to the whole image; object detection locates and classifies multiple objects with bounding boxes
Image recognition (classification) outputs one label per image, while object detection provides bounding boxes and labels for all object instances found in the image.