CORTEX.LAB / INTERACTIVE CV LABORATORY / V0.4

Computer vision,
made visible.
read about it. play with it.

A live laboratory for convolutions, kernels, embeddings & neural architectures. Every concept is a slider, every tensor a shape you can rotate, every gradient an animation. No tutorials. No videos. Just experiments.

START WITH CONVOLUTIONS → OPEN CNN BUILDER 20 MODULES · 85+ EXPERIMENTS
↹ traverse modules ⌘ K command palette Space play/pause sims R randomize
20
Modules
85
Live Experiments
∞
Tensor States
0
Server Calls
// 01

The Modules

20 / 20 ONLINE

Each module is a self-contained interactive notebook. Drag, scrub, click anything. Every visualization runs locally in your browser — real math, no canned animations.

M01 · FOUNDATIONS
↗

Matrix & Tensor Playground

Rotate, multiply, broadcast. Watch shapes cascade through dimension flips and dot products in real time.

M02 · FOUNDATIONS
↗

Convolution Kernels

Scrub a 3×3 kernel across procedural imagery. Sobel, Laplacian, custom — see every multiply & accumulate.

M03 · FOUNDATIONS
↗

Feature Extraction

From edges to eyes. Walk the hierarchy of representations from layer 1 through layer 18.

M04 · FOUNDATIONS
↗

Activation Functions

The non-linear bend that makes deep nets work. Plot f(x) and its gradient; compare ReLU, GELU, sigmoid & more.

M05 · FOUNDATIONS
↗

Loss Functions

The number training shrinks. MSE vs MAE vs Huber; cross-entropy vs focal. Drag the error, watch the penalty.

M06 · FOUNDATIONS
↗

Optimizers

SGD, Momentum, Adam & friends race down bowls, ravines and Rosenbrock. See who reaches the bottom first.

M07 · FOUNDATIONS
↗

Regularization

Stop a model memorizing. Live overfitting demo, L1 vs L2 geometry, dropout, and the full toolbox.

M08 · FOUNDATIONS
↗

The MLP · step-by-step

One full training step with every number shown: forward, loss, backprop chain rule, weight update. The atom of deep learning.

M09 · ARCHITECTURES
↗

CNN Architecture Builder

Drag layers in, watch tensor shapes propagate. Compare your design against ResNet, VGG, MobileNet.

M10 · ARCHITECTURES
↗

Attention & Transformers

From RNNs to Q/K/V to ViT. The full block anatomy, then hover patches to watch them attend.

M11 · ARCHITECTURES
↗

CLIP & OpenCLIP

Two encoders, one space. Watch the contrastive matrix train, then classify zero-shot over six classes.

M12 · ARCHITECTURES
↗

DINO · Self-Supervised

No labels. Student matches teacher across crops — and attention maps segment objects on their own.

M13 · ARCHITECTURES
↗

GANs

A forger and a detective in a duel. A real 2D adversarial game — train it live, watch mode collapse.

M14 · LABORATORY
↗

OpenCV Lab

Threshold, contour, morph. The classical CV toolbox as one big set of sliders.

M15 · LABORATORY
↗

Augmentation Lab

Flips, jitters, cutout, mixup. Build an augmentation pipeline and preview a batch.

M16 · LABORATORY
↗

Training Simulator

Watch loss curves descend, gradients propagate, decision boundaries form. Tune LR, optimizer, epochs.

M17 · LABORATORY
↗

Training MLP

Watch a network learn neuron by neuron — live weights, activations, gradient pulses and a forming boundary.

M18 · LABORATORY
↗

Representation Learning

The engine under CLIP & DINO, with normal backbones. Contrastive, triplet, emergent clusters.

M19 · LABORATORY
↗

Embeddings & Latents

Fly through 3D latent space. Cluster, interpolate, find nearest neighbors. PCA / t-SNE / UMAP.

M20 · LABORATORY · FLAGSHIP
↗

AI Sandbox  ·  node editor

Chain everything. A visual node graph where the output of any module becomes the input of any other. Build pipelines that ingest an image, push it through edges → features → attention → embedding, and visualize every intermediate tensor.

// 02

How it works

01
Pick a concept.
Pick any module. Each one stands alone — start anywhere.
02
Manipulate the inputs.
Drag sliders, modify matrices, add layers, swap activations. The visualization recomputes on every frame.
03
Read the math.
Equations live alongside every visualization. Toggle them on/off, hover for derivations.
04
Chain it together in the Sandbox.
When you're ready, drop modules onto a canvas and connect them. Build a real CV pipeline.
// 03

What's inside

SYSTEM MANIFEST 20 / 20 MODULES
FOUNDATIONS8 modules
↳ matrix.ops12 demos
↳ convolution.live8 demos
↳ feature.hierarchy5 demos
↳ activation.plot9 demos
↳ loss.plot8 demos
↳ optimizer.race6 demos
↳ regularization7 demos
↳ mlp.trace8 steps
ARCHITECTURES5 modules
↳ cnn.builder9 demos
↳ attention.map6 demos
↳ clip.contrastive5 demos
↳ dino.selfsup5 demos
↳ gan.adversarial6 demos
LABORATORY7 modules
↳ opencv.toolbox11 demos
↳ augmentation.pipe7 demos
↳ training.sim6 demos
↳ training.mlplive
↳ representation.lab4 demos
↳ latent.space5 demos
↳ sandbox.graph∞ demos
For students, not consumers. No paywalls, no accounts, no telemetry. Every page is one HTML file. View source, fork it, break it.
CORTEX.LAB · CV INTERACTIVE
BUILD 0.4.1 · COMMIT a3f721c · 16 MAY 2026
WEBGL 2.0 CANVAS 2D TYPED ARRAYS OFFSCREEN.CANVAS 100% LOCAL