🖼️ Image Data in ML

How does a computer actually "see" an image?

🔢 The Pixel Grid

To a computer, an image is just a massive 2D matrix of numbers.

• A grayscale image has one channel (numbers from 0 to 255 representing brightness).
• A color image has three channels: Red, Green, and Blue (RGB). So it's essentially three matrices stacked like pancakes!

🐍 Python Implementation

import torch

# A tiny 3x3 Grayscale image
grayscale_img = torch.tensor([
    [255, 255, 255], # White
    [0,   0,   0  ], # Black
    [128, 128, 128]  # Gray
])

# PyTorch prefers image shapes to be: (Channels, Height, Width)
# Let's create a fake 64x64 RGB image
color_img = torch.rand(3, 64, 64) # 3 channels (RGB)
print("Image Shape:", color_img.shape)