Hey — Steve here.

Last week we covered NLP. This week, let's talk about something you've probably heard but maybe never fully understood.

This Week's Term: Computer Vision

The quick version: Teaching computers to understand images and video. Like showing a toddler 10 million cat photos until they can spot a cat anywhere — except the toddler is a GPU.

The deeper version:

You glance at a photo and instantly know there's a dog sitting on a couch in a living room. You didn't think about it. Your brain just... did it. Computers had to be painstakingly taught to do what your eyes and brain do automatically.

Computer vision is the field of AI that gives computers the ability to "see" and understand images and video. Not just detect pixels — actually understand what's in the picture. "That's a face." "That's a stop sign." "That car is about to hit that other car." It's what powers everything from Face ID on your phone to self-driving cars to those Instagram filters that put dog ears on your head.

The wild part? Modern computer vision systems can sometimes see things humans miss. They can spot tumors in X-rays that doctors overlooked, find defects in manufactured parts, or read license plates at highway speeds. They went from barely recognizing a cat to outperforming human experts in certain narrow tasks.

Why This Matters

Because computer vision is increasingly watching you — literally. It's in your phone's camera, your doorbell camera, security systems, retail stores, and your car. Understanding it helps you understand both the amazing possibilities (medical diagnosis, accessibility tools) and the legitimate concerns (surveillance, deepfakes, bias in facial recognition).

Try It Yourself (2 minutes)

Open Google Lens on your phone and point it at objects. It identifies what it sees in real time — that's computer vision.

Go Deeper

• Search for "Computer Vision explained simply" — you'll find great visual guides
• 🔗 Read the full SpeakNerd term page

The Nerd Corner

Computer vision uses convolutional neural networks (CNNs), vision transformers (ViTs), and other deep learning architectures to extract features from visual data. Key tasks include image classification, object detection (YOLO, R-CNN), semantic segmentation, pose estimation, and optical character recognition (OCR). Modern systems leverage transfer learning from large pre-trained models like CLIP for zero-shot visual understanding.

Next week: Multimodal — we'll break down what it means and why you should care.

— Steve

P.S. Know someone who'd find this useful? Forward this email. They can sign up here.