Color spaces (RGB, BGR, grayscale, HSV) in Computer Vision - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Predict Output

intermediate

What is the output shape after converting an RGB image to grayscale using OpenCV?

Given an RGB image with shape (100, 150, 3), what will be the shape of the image after converting it to grayscale using OpenCV's cvtColor function?

Computer Vision

import cv2
import numpy as np

image_rgb = np.random.randint(0, 256, (100, 150, 3), dtype=np.uint8)
gray_image = cv2.cvtColor(image_rgb, cv2.COLOR_RGB2GRAY)
print(gray_image.shape)

A (100, 150)

B (150, 100)

C (100, 150, 1)

D (100, 150, 3)

💡 Hint

Grayscale images have only one channel, so the shape loses the last dimension.

When converting an RGB image (height, width, 3) to grayscale, the output image has shape (height, width) because grayscale images have only one channel.

🧠 Conceptual

intermediate

Why does OpenCV use BGR instead of RGB by default?

OpenCV uses BGR color order by default instead of RGB. What is the main reason for this?

A Because BGR matches the order used by Windows bitmap files historically.

B Because BGR is more efficient for GPU processing.

C Because RGB is patented and requires licensing fees.

D Because BGR uses less memory than RGB.

💡 Hint

Think about historical image file formats and compatibility.

OpenCV uses BGR by default because it matches the color order used in Windows bitmap files, which historically stored pixels in BGR order.

❓ Metrics

advanced

Which metric is best to compare color similarity in HSV space?

You want to measure how similar two colors are in HSV space. Which metric is most appropriate?

A Cosine similarity on grayscale intensities

B Manhattan distance on RGB values

C Euclidean distance on HSV values

D Hamming distance on binary color codes

💡 Hint

Consider the numeric nature of HSV and how distances reflect similarity.

Euclidean distance on HSV values captures differences in hue, saturation, and value, making it suitable for color similarity in HSV space.

🔧 Debug

advanced

Why does this code produce a blueish image instead of red?

You load an image using OpenCV and try to display a pure red color by setting pixel values to (255, 0, 0). The displayed image looks blueish. Why?

Computer Vision

import cv2
import numpy as np

image = np.zeros((100, 100, 3), dtype=np.uint8)
image[:] = (255, 0, 0)  # Set to red
cv2.imshow('Image', image)
cv2.waitKey(0)
cv2.destroyAllWindows()

A The data type uint8 causes color distortion.

B OpenCV uses BGR order, so (255, 0, 0) is blue, not red.

C The display window is not refreshing properly.

D The image array is not initialized correctly and contains random values.

💡 Hint

Check the color channel order OpenCV expects.

OpenCV uses BGR color order by default, so setting (255, 0, 0) sets the blue channel to max, producing a blue image instead of red.

❓ Model Choice

expert

Which color space is best for skin tone detection in images?

You want to build a model to detect skin tones in images robustly under different lighting. Which color space is generally best to use as input features?

A RGB, because it is the raw color data from cameras.

B Grayscale, because it simplifies the image to intensity only.

C BGR, because OpenCV uses it by default.

D HSV, because it separates color information from brightness.

💡 Hint

Think about separating color from light intensity to handle lighting changes.

HSV separates hue and saturation from brightness, making it easier to detect skin tones under varying lighting conditions.