Question 1

What are the 5 stages of the Computer Vision process in Unit 3?

Accepted Answer

The five sequential stages are: 1) Image Acquisition (capturing via digital cameras, scanners, MRI/CT); 2) Preprocessing (noise reduction, normalization 0-1, resizing, histogram equalization); 3) Feature Extraction (edge detection, corner detection, texture analysis, color features); 4) Detection/Segmentation (classification, localization, multi-object detection, semantic vs instance segmentation); and 5) High-Level Processing (scene understanding, context reasoning, and decision making).

Question 2

What is the difference between Classification and Object Detection?

Accepted Answer

Classification evaluates an image as a whole and assigns a single class label to it. Object Detection identifies, locates (draws bounding boxes around), and classifies multiple objects across different classes in the same image using algorithms like R-CNN, R-FCN, YOLO, and SSD.

Question 3

What is the difference between Semantic Segmentation and Instance Segmentation?

Accepted Answer

Semantic Segmentation classifies each pixel belonging to a category without differentiating individual objects of that same category (e.g. all sheep are masked in the same color). Instance Segmentation differentiates every individual instance of the same class with a unique pixel-wise mask (e.g. Sheep 1, Sheep 2, and Sheep 3 each receive separate masks).

Question 4

How does a computer represent digital grayscale and RGB images?

Accepted Answer

A computer interprets images as 2D grids of pixels. In monochrome/grayscale images, each pixel has an intensity value from 0 (black) to 255 (white). In RGB color images, each pixel consists of 3 channels (Red, Green, Blue) each having 8 bits (1 byte = 256 values), resulting in over 16.7 million possible color combinations (256 x 256 x 256).

Question 5

What OpenCV functions are taught for Python image processing?

Accepted Answer

The curriculum introduces: cv2.imread('path') to load an image, cv2.imshow('title', img) to display it, cv2.waitKey(0) and cv2.destroyAllWindows() to manage the viewer window, cv2.resize(img, (width, height)) to standardize image dimensions, and cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) to convert color images to grayscale.

Making Machines SeeComputer Vision Architecture & Pipelines

3.1 & 3.2 How Machines See: Digital Representation & Pixels

Monochrome & Grayscale Images

The RGB Color Model (16M Colors)

Next: Unit 4 · AI with Orange Data Mining Tool