OpenCV · Python · browser lab
Every concept runs in the browser. Write Python, watch the image change, and learn to see the way the algorithm does.
LLMs can describe a photo in words, but the answers can be vague or inconsistent. OpenCV gives you exact numbers from the pixels: locations, counts, and timing, and the same code produces the same result every run.
No environment setup. No downloading datasets. Every lesson is bite-sized and interactive, so you can make real progress in short bursts and always know what to expect from your output.
What you will learn
From pixels and crops through convolution, edges, and image geometry.
This lab is still growing. More lessons and phases are on the way.
Treat bitmaps as measurable arrays: indexing, channels, crops, and resizing without swapping axes.
Convolution as weighted neighborhoods: kernels, filter2D, gradients, and Canny as a staged detector.
From BGR to HSV: why color spaces matter, then lighting-tolerant segmentation with global and adaptive thresholds, morphological cleanup, and bitwise masking.
2D transforms as matrix multiply: rotation, scale, translation with warpAffine, then perspective and homography for planar scenes.
From binary regions to boundaries: findContours, hierarchy, area filtering, and simple shape metrics.