juniorCNN
What is a feature map in CNNs?
Updated May 15, 2026
Short answer
A feature map is the output produced by applying a filter to an input image.
Deep explanation
Each convolution filter produces a feature map representing where a specific pattern appears in the input. Early layers detect edges, while deeper layers detect complex structures like objects.
Real-world example
Detecting road lanes in autonomous driving systems.
Common mistakes
- Thinking feature maps are manually designed features.
Follow-up questions
- Why do deeper layers have more abstract feature maps?
- Can feature maps be visualized?