Computer Vision Overview — ML Technical Interview Questions & Answers

Teaching machines to interpret images and video.

Computer Vision Overview

  • Simple meaning: Computer vision = models on images/video: detect a face, read a digit, find a tumour blob. Pixels are a grid — CNNs (and now vision transformers) fit that shape.
  • Picture: 28×28 grayscale digit → CNN → 0–9. Data needs labels (or self-supervised tricks). Augmentation (flip, crop) grows the sheet.
  • Fresher: MNIST/CIFAR with a small CNN. Know IoU for detection boxes if they ask beyond classify.
  • Trap: training on 30 photos and claiming production vision. Trap: no normalisation of pixels (0–255 vs 0–1).

Technical Interview Tip: Images are grids. CNN classifies. Augment and scale pixels.

Code Examples for Computer Vision Overview

Computer Vision Overview — working example

# PCA-style intuition: project 2D → 1D by a direction vector
points = [(1.0, 1.0), (2.0, 2.0), (3.0, 2.8)]
# unit direction along y = x
ux, uy = 0.7071, 0.7071
proj = [x * ux + y * uy for x, y in points]
print([round(v, 3) for v in proj])

# Tiny ensemble: average two model votes
votes = [1, 0, 1]
v2 = [1, 1, 1]
ens = [1 if a + b >= 2 else 0 for a, b in zip(v1, v2)]
print(ens)

Computer Vision Overview — dim reduction compresses features; ensembles combine models; deep nets/RL/NLP are specialized families — practice the core math idea first.

How to Explain Computer Vision Overview to an Interviewer

Typical CV model?

CNN or vision transformer

Augmentation?

Flip/crop so the model sees more variety