Computer Vision

Create a free account to open this chapter.

• Computer vision, image processing, machine learning, and deep learning ~18 min
• The perception pipeline, and how computer vision got here ~18 min
• Image understanding vs. generation: what does a computer actually see? ~15 min

Create a free account to open this chapter.

• Sensors, lenses, and exposure ~16 min
• Sampling and aliasing at the sensor ~16 min
• Bayer filters, demosaicing, and a first pass at calibration ~16 min

Create a free account to open this chapter.

• Tracing and describing a shape's boundary ~14 min
• Moments and Hu moments ~16 min
• Contour approximation ~13 min

Create a free account to open this chapter.

• What makes a good keypoint ~14 min
• Corner detection: Harris, Shi-Tomasi, and FAST ~24 min
• SIFT, ORB, and BRIEF descriptors ~24 min

Create a free account to open this chapter.

• Matching descriptors across two images ~16 min
• RANSAC as a general robust-fitting idea ~16 min
• The feature-matching recipe ~16 min

Create a free account to open this chapter.

• Homogeneous coordinates ~15 min
• Affine vs. projective transforms ~16 min
• Homography ~18 min

Create a free account to open this chapter.

• The pinhole model and the camera matrix ~22 min
• Lens distortion and calibration ~20 min
• The camera calibration recipe ~18 min

Create a free account to open this chapter.

• Epipolar geometry ~20 min
• Disparity and triangulation ~22 min
• Dense vs. sparse correspondence, and point clouds ~20 min

Create a free account to open this chapter.

• Pose estimation and two-frame structure from motion ~18 min
• Multi-frame structure from motion and bundle adjustment ~18 min
• Visual SLAM and loop closure ~18 min

Create a free account to open this chapter.

• Turning an image into a feature vector ~18 min
• Classifying a feature vector: k-NN, SVM, and random forests ~22 min
• K-means and PCA for vision ~20 min
• Classical features vs. a neural network: the honest comparison ~16 min

Create a free account to open this chapter.

• Forward pass, loss, and gradients: for a vision task ~16 min
• Convolution as a learned operation, pooling, and receptive fields ~26 min
• Training discipline: splits, augmentation, and overfitting ~22 min
• Transfer learning and fine-tuning ~24 min

Create a free account to open this chapter.

• CNN architecture families, a survey ~20 min
• Patches, attention, and Vision Transformers ~20 min
• Choosing a backbone: the decision framework ~22 min

Create a free account to open this chapter.

• Bounding boxes, IoU, and confidence ~16 min
• Non-max suppression ~15 min
• One-stage, two-stage, and transformer detectors ~14 min
• Precision, recall, and mAP ~18 min

Create a free account to open this chapter.

• Semantic vs. instance segmentation ~19 min
• FCN, U-Net, and Mask R-CNN ~21 min
• Dice and IoU for segmentation ~17 min

Create a free account to open this chapter.

• Point-, box-, and text-prompted segmentation ~18 min
• Segmenting every instance of a concept ~16 min
• From task-specific training to promptable foundation models ~14 min

Create a free account to open this chapter.

• Latent spaces and image embeddings ~16 min
• Cosine similarity and nearest-neighbor search ~20 min
• Retrieval pipelines and why generic embeddings win ~18 min

Create a free account to open this chapter.

• Contrastive image-text embedding ~20 min
• Zero-shot classification and retrieval ~22 min
• Multimodal assistants: captioning and VQA ~14 min

Create a free account to open this chapter.

• The OCR pipeline: detection vs. recognition ~17 min
• Document preprocessing and layout analysis ~18 min
• The complete pipeline, worked end to end ~19 min

Create a free account to open this chapter.

• Optical flow and background subtraction ~22 min
• Kalman-filter tracking ~20 min
• Multi-object tracking and occlusion ~22 min

Create a free account to open this chapter.

• 2D/3D pose keypoints and skeletons ~24 min
• Monocular depth estimation ~20 min
• Neural scene representation: NeRF vs. 3D Gaussian Splatting ~20 min

Create a free account to open this chapter.

• Autoencoders and VAEs ~12 min
• GANs ~11 min
• Diffusion models ~11 min

Create a free account to open this chapter.

• Cross-task evaluation vocabulary ~20 min
• Quantization and ONNX export ~18 min
• The complete pipeline, and capstone projects ~22 min

Create a free account to open this chapter.

• Colab and Kaggle: free GPU notebooks ~9 min
• Hugging Face Spaces and your first Gradio demo ~9 min
• Other cloud options: a quick comparison ~7 min

Create a free account to open this chapter.

• Model cards, dataset cards, and loading a pretrained model ~14 min
• Inference vs. fine-tuning, and choosing a model for your hardware ~14 min
• Quantization and ONNX export in practice ~15 min

Create a free account to open this chapter.

• Detection and segmentation -- current practice ~8 min
• Embeddings and vision-language models -- current practice ~8 min
• Applied recognition and 3D scene representation -- current practice ~9 min