English | Türkçe

Image Processing and Computer Vision Guide

A comprehensive, practical guide to building computer vision applications with OpenCV and modern deep learning frameworks.

This guide bridges theory and practice for both students and working developers. Each chapter opens with a real-world problem, builds intuition before introducing formulas, and delivers complete, copy-paste-ready Python examples. Students learn the concepts; developers ship the code.

31 chapters · Python 3.10+ · OpenCV 4.9+ · PyTorch · Ultralytics YOLOv8

---

Table of Contents

• Documentation
• Getting Started
• Example Projects
• Contributing
• Support the Project
• Release History

---

Documentation

Every chapter follows the same structure: real-world problem → intuitive explanation → formula (where needed) → complete working Python code → comparison table → summary and references.

Chapter	Summary
What is OpenCV?	OpenCV history and architecture, module overview, comparison with alternative libraries, OpenCV 4.x/5.0 highlights.
Development Environment	Setting up Python, VS Code, Jupyter Notebook, Google Colab, and Conda for computer vision development.
OpenCV Wrappers	EmguCV, JavaCV, LiveCV, RubyCV — what they are and when to use them.
Installation and Compilation	OpenCV installation for Windows, Linux, macOS, and Raspberry Pi.
IDE Configuration	Configuration for VS Code, PyCharm, Conda, Google Colab, and Android Studio.
Introduction and Basic Concepts	Digital image as a matrix, BGR/RGB difference, reading from file and camera, pixel manipulation.
Video Recorders and Codecs	Codec, FourCC, and video recording with cv2.VideoWriter.
Image Manipulation	Resize, rotate, affine and perspective transforms; document scanner scenario.
Color Spaces and Histogram	BGR, HSV, Lab color spaces; color-mask object detection; histogram equalization and CLAHE.
Morphological Image Processing	Erosion, dilation, opening, closing, morphological gradient; Otsu and adaptive thresholding.
Filters and Edge Detection	Convolution intuition, Gaussian/Median/Bilateral blur, Sobel, four-stage Canny pipeline.
Background Subtraction	Frame differencing, MOG2 Gaussian mixture model, KNN subtractor; security camera scenario.
Video Analysis and Object Tracking	Optical flow constraint, Lucas-Kanade pyramidal LK, Farneback dense flow, CSRT and KCF trackers.
Object Detection and Recognition	Haar Cascade, Template Matching, real-time YOLOv8 detection; IoU, Precision/Recall, mAP metrics.
Camera Calibration and 3D Vision	Pinhole camera model, radial distortion correction, chessboard calibration, stereo depth map.
Face Recognition	Eigenfaces/LBPH, FaceNet embedding intuition, ArcFace angular margin, DeepFace integration.
Optical Character Recognition (OCR)	OCR pipeline, image preprocessing, Tesseract PSM modes, EasyOCR, EAST text detection.
Feature Extraction and Matching	SIFT DoG scale space, ORB, FLANN matching, Lowe ratio test, RANSAC homography; image stitching.
GPU and Parallel Computing	CUDA module, OpenVINO model acceleration, threading for parallel image processing.
Pose Estimation	17 COCO keypoints, MediaPipe Pose skeleton detection, angle calculation, rep counter, YOLOv8-Pose.
Segmentation	Semantic/instance/panoptic differences, Watershed, YOLOv8-seg, SAM (Segment Anything), IoU and mIoU.
Edge Deployment	Model optimization with ONNX Runtime and TFLite, Raspberry Pi and mobile deployment.
OpenCV Mobile	OpenCV on Android and iOS, camera stream processing, TFLite integration.
Augmented Reality	ArUco marker detection, 3D axis projection, image overlay, MediaPipe face filter.
Vision Transformers	Attention mechanism intuition, ViT patch embedding and CLS token, Swin shifted window, DETR, CNN vs ViT.
Generative Models and Diffusion	GAN minimax game intuition, VAE latent space, DDPM noise process, Stable Diffusion, ControlNet.
3D Vision	Stereo and monocular depth estimation, Open3D point cloud, PointNet permutation invariance, NeRF.
Video Understanding and Action Recognition	Two-Stream, SlowFast dual-pathway architecture, VideoMAE tube masking, HuggingFace action classification.
Model Training and Evaluation	Data preparation, augmentation, Mixup, ResNet50 fine-tuning, Cosine Annealing, confusion matrix.
Vision-Language Models	CLIP contrastive loss, zero-shot classification, LLaVA, BLIP-2 Q-Former; visual question answering.

---

Getting Started

Install the core dependencies:

pip install opencv-python numpy matplotlib

For deep learning and modern chapters:

pip install torch torchvision ultralytics mediapipe easyocr deepface transformers diffusers open3d

All code examples require Python 3.10+ and OpenCV 4.9+. Sample source code is available in the /code directory.

---

Example Projects

Topic	Python	Java	C++	C#	JavaScript
Video I/O	Video and Camera I/O Video Recorder
Image Manipulation	Geometric Shape Drawing Image Crop Pixel Manipulation Image Resize
Color Spaces	Color Space Conversion
Morphological Image Processing	Opening Closing Dilation Gradient Erosion
Filters	Gaussian Sobel Laplacian & Sobel
Edge Detection	Canny GaussianBlur & Canny & Sobel & Prewitt				Canny
Background Subtraction	Absdiff MOG / MOG2 / GMG
Video Analysis and Object Tracking	MeanShift CamShift	Color-Based Object Tracking
Object Detection and Recognition	Template Matching Object Recognition with DNN Face Detection with Haar Cascade	Template Matching Object Recognition with DNN Face Detection with Haar Cascade
Feature Extraction
Camera Calibration and 3D Vision	Fisheye Lens Correction Camera Calibration
Machine Learning and Deep Learning	Face Detection with DNN	Gender Classification with DNN
Face Recognition	Face Recognition with face_recognition Face Recognition with FaceNet			Face Recognition with Eigenfaces and Fisherfaces
Optical Character Recognition (OCR)	OCR with Tesseract
Augmented Reality	Hat Filter — Face Detection and Hat Overlay

---

Contributing

Contributions are welcome — bug reports, new chapter proposals, code corrections, or translations. See CONTRIBUTING.en-US.md for guidelines.

Questions and discussions go in Issues. For direct contact, reach out via the website or email.

---

Support the Project

This guide is a free, open-source educational resource maintained entirely on a volunteer basis. If it has been useful to you and you'd like to support new chapters, updated examples, and the continued growth of the community, consider sponsoring via GitHub Sponsors.

Your support directly funds new content, code maintenance, and the broader Turkish computer vision community.

---

Release History

v3.0 — May 2026

The most comprehensive update since the repository's launch in 2018. Migration to a Python-first stack, eleven new chapters, and a ground-up rewrite of all existing chapters happened simultaneously. An O'Reilly-style pedagogical structure was adopted throughout: each topic opens with a real-world problem, intuition precedes formulas, and every code example is complete and runnable.

Rewritten chapters (15): 1, 6, 8, 9, 10, 11, 12, 13, 14, 15, 17, 18, 19, 21, 22

New chapters (11):

• Chapter 20: GPU and Parallel Computing — CUDA, OpenVINO, Threading
• Chapter 21: Pose Estimation — MediaPipe Pose, YOLOv8-Pose, angle calculation
• Chapter 22: Segmentation — YOLOv8-seg, SAM, DeepLab
• Chapter 23: Edge Deployment — ONNX Runtime, TFLite, quantization
• Chapter 24: OpenCV Mobile — Android, iOS
• Chapter 25: Augmented Reality — ArUco, 3D Overlay, Face Filter
• Chapter 26: Vision Transformers — ViT, Swin Transformer, DETR
• Chapter 27: Generative Models and Diffusion — GAN, VAE, DDPM, Stable Diffusion
• Chapter 28: 3D Vision — PointNet, NeRF, Depth Estimation
• Chapter 29: Video Understanding and Action Recognition — SlowFast, VideoMAE
• Chapter 30: Model Training and Evaluation — Transfer Learning, Augmentation
• Chapter 31: Vision-Language Models — CLIP, LLaVA, BLIP-2

---

v1.0 — 2018–2020 · v1 tag

Initial release. Java, Python, and C++ examples with OpenCV 3.x/4.x. Fundamental image processing topics.