Computer VisionLesson 2
Lesson 28 min
Filters & Convolution
The core operation that detects edges and patterns.
What you will learn
- ✓What a convolution kernel is
- ✓How filters detect features
- ✓Why this is the building block of CNNs
Explanation
A filter (or kernel) is a small grid of numbers that slides across the image, multiplying and summing as it goes — this is convolution.
Different kernels detect different things: one finds vertical edges, another blurs, another sharpens. The output highlights wherever that pattern appears.
Classic vision hand-designed these filters; modern deep learning learns them from data instead.
Real-world use
The 'edge detect' and 'sharpen' options in photo editors are literally convolution filters.
Common mistakes
- • Thinking the network is hand-programmed with filters — in CNNs the filters are learned.
Practice
Describe what an edge-detection filter would highlight in a photo of a building.
Knowledge check
0/1 answered1. What does a convolution do?
Answer all questions to check.