47. Convolutional Neural Networks (CNNs) for Images

47.4 CNN Stack Overview

By Ravindra Bagale · Last updated: 24 October 2026

Why. Real CNNs stack conv → activation → pool (repeated), then flatten into a small classifier head.

What. Pattern: extract features with conv blocks, then nn.Linear layers for class scores (pedha vs chips on a fictional shelf).

How.

import torch.nn as nn

class TinyCNN(nn.Module):
 def __init__(self):
 super().__init__()
 self.features = nn.Sequential(
 nn.Conv2d(1, 8, 3, padding=1), nn.ReLU(), nn.MaxPool2d(2),
 nn.Conv2d(8, 16, 3, padding=1), nn.ReLU(), nn.MaxPool2d(2),
 )
 self.head = nn.Linear(16 * 7 * 7, 2) # for 28x28 input
 def forward(self, x):
 x = self.features(x)
 x = x.view(x.size(0), -1)
 return self.head(x)

print(TinyCNN())