Cheat Sheets

Dense, one-page references to keep open while you work. Each one summarizes chapters that teach the material in depth.

Data and classical vision

• OpenCV and image I/O

  ---

  Reading, writing, color conversions, resizing, drawing, video, and the pitfalls table.

• Image processing

  ---

  Histograms, thresholds, filters, edges, frequency, morphology, contours, and warps.

• Geometry and cameras

  ---

  Transforms, camera models, calibration, epipolar geometry, stereo, and RANSAC.

Deep learning for vision

• torchvision and augmentation

  ---

  Datasets, transforms v2, model zoos, and training recipes.

• Object detection

  ---

  Box formats, IoU, NMS, mAP, and detector families.

• Segmentation

  ---

  Losses, metrics, mask formats, and architectures.

Modern vision

• Vision transformers and VLMs

  ---

  ViT anatomy, self-supervised methods, CLIP, and zero-shot recipes.

• Diffusion models

  ---

  The forward and reverse processes, samplers, guidance, and latent diffusion.

• Deploying vision models

  ---

  ONNX, quantization, runtimes, and latency budgets.

Reference

• Glossary

  ---

  Every term in the handbook, defined in one place.