Home▸Computer Science▸CNN Vision Lab: Convolution, Detection & Segmentation

🧠 CNN Vision Lab: Convolution, Detection & Segmentation

Feed a toy image through real 2D convolution + ReLU + max-pooling stages and switch between a Classification, Object-Detection (YOLO-style) and Segmentation head to see how one CNN backbone feeds three different vision tasks.

Computer Science3DAdvanced60 FPS

⚙ Under the hood

Feed a toy image (circle, square, triangle, cross) through real 2D convolution + ReLU + max-pooling stages and watch the feature maps shrink layer by layer. Switch between a Classification head (softmax over shape templates), an Object Detection head (YOLO-style grid + bounding box) and a Segmentation head (per-pixel mask from the deepest feature map) to see how the same CNN backbone feeds three different task heads.

Three.jsCNNComputer VisionConvolutionObject DetectionYOLOSegmentationVision Transformer

3D · Three.js / WebGL renderer · 60 FPS target · runs fully client-side, no install