Receptive Field
A single neuron in a CNN only looks at a tiny window of the image — so how does the network ever "see" a whole bus? The answer is the receptive field: the region of the input that can influence a neuron. Stacking small convolutions widens that window layer by layer, and this topic derives the math, exposes the effective-receptive-field gap, and connects to dilated and multi-scale designs.
01.The Problem: One Neuron Sees Only Nine Pixels
Picture a photo of a street. A bus fills half of it.
Inside a convolutional network (a CNN — a network whose layers slide small filters over the image), the very first layer uses 3x3 kernels. Each neuron there multiplies its weights against 9 pixels. Just 9.
So the question becomes
How does a neuron that only sees 9 pixels ever know there is a whole bus in the image?
And deeper in the network,
Which input pixels can actually change this neuron's output?
A single neuron never sees the whole image. But its descendants combine many local views, and the views themselves combine many local views. Information travels upward in chains.
The set of input pixels at the start of those chains has a name: the receptive field.
Getting it right is an architecture decision. If your final neurons can't "see" enough of the image, no amount of training will let them classify a large object — the pixels simply never reach them.
Receptive field expands with depth 🪟
Receptive field expands with depth 🪟
Each stacked 3x3 (stride 1) layer widens the receptive field by roughly 2 pixels. By the final layers of a deep network, a single neuron effectively sees a large fraction of the input image.
Unlock Topic #102: Receptive Field
You are viewing a preview. The full in-depth technical walkthrough, worked derivations, and code notebooks for this concept, along with self-assessment quizzes, are available with Pro or Lifetime Access.
Failure modes, high-throughput bottlenecks, and real FAANG implementation decisions.
Interactive system topology diagrams, live parameter simulators, and downloadable SVG charts.
Staff-level multiple-choice quiz questions with instant feedback and answer explanations.
Firebase Google authentication automatically syncs your completed topics and quiz scores.
How clear and actionable was this distributed systems breakdown?