Digital image
A digital image is visual content encoded as data, typically as a two-dimensional arrangement of pixels or equivalent visual elements.
Deep Learning with Python, Third Edition
Data in which each example is a 2D image with one or more color channels, typically stored as a rank-3 tensor per sample.
- How it is used
- Supplies pixel-grid inputs for image models and adjustable inputs for visualizing convolutional filters.
- How deep it goes
- Defines image tensor axes and demonstrates flattening and scaling, without detailing image-file storage.
- Examples
- Uses flattened handwritten-digit images, RGB versus HSV color encoding, and blank images optimized to activate filters.
- What you can do after
- Apply image reshaping and scaling in a NumPy and Keras workflow, and understand why pretrained models need matching preprocessing.
Read how Deep Learning with Python, Third Edition explains digital image →
Learn Haskell by Example
An image file is a digital file containing encoded visual data such as pixels or vector descriptions.
- How it is used
- Models PNM files for parallel image processing and later PNG export.
- How deep it goes
- Builds a typed image representation, explains Vector trade-offs, and implements conversion to JuicyPixels output values.
- Examples
- Uses grayscale and RGB PNM variants at 8- and 16-bit depth, then converts them to PNG images.
- What you can do after
- Build it from scratch as a typed PNM image representation and export it through JuicyPixels as PNG.
Data Analysis with LLMs
Image data is data represented as visual images containing depicted objects, people, scenes, or other visual information.
- How it is used
- Supports multimodal analysis by supplying visual data for questions, comparisons, tagging, and other application workflows.
- How deep it goes
- Explains files, PNG filtering, base64 encoding, dictionaries, pairwise comparison, prompts, and output naming without discussing image internals.
- Examples
- Uses PNG files, reference portraits, beach.png, filename prefixes, base64 strings, and Yes-or-No comparison prompts.
- What you can do after
- Build image-processing applications with Python, GPT-4o, multimodal prompts, local directories, and the requests API.
Causal AI
An image is a two-dimensional arrangement of visual measurements, such as pixel intensities, representing a scene, object, or symbol.
- How it is used
- Represents visual observations as model inputs and as outcomes generated from interpretable causal factors.
- How deep it goes
- Defines pixel matrices and develops encoder-decoder counterfactual generation, while treating individual-pixel causality as an unsuitable abstraction.
- Examples
- Uses MNIST and TMNIST digit matrices plus dSprites shapes varied by shape, scale, rotation, and position.
- What you can do after
- Apply it via PyTorch encoders and decoders to reconstruct images and generate images under changed causal factors.
WordPress Power Toolkit: Harness AI to build next-level websites
An image is a visual digital asset displayed as part of online content.
- How it is used
- Uses images to strengthen website content and give visitors reasons to stay, while avoiding excessive or merely decorative media.
- How deep it goes
- Explains relevance, contrast, placement, size, context, style consistency, and page-load testing.
- Examples
- Uses recipe home pages, leaves, raw potatoes, cooking imagery, and bass-fishing image counts.
- What you can do after
- Diagnose image choices and page-load problems using design checks and pagespeed.web.dev.
Read how WordPress Power Toolkit: Harness AI to build next-level websites explains digital image →
Build a Website with ChatGPT
A file containing encoded image data that can be displayed on a web page.
- How it is used
- Uses images to enliven web pages, supplement text, convey information concisely, and support reader retention.
- How deep it goes
- Explains separate-file storage, image formats, HTML references, accessibility text, thumbnails, captions, and lightboxes.
- Examples
- Uses book-club logos, book covers, online-journal logos, and full-window hero images as concrete web-page imagery.
- What you can do after
- Apply it via HTML by storing an image file, referencing it with img and src, and describing it with alt text.
Read how Build a Website with ChatGPT explains digital image →
Bayesian Optimization in Action
Image data is numerical or symbolic data representing visual content, typically as arrays of pixel values.
- How it is used
- Represents handwritten digits as tensor features for visualization, Gaussian-process modeling, and image-data experiments.
- How deep it goes
- Defines pixels, tensor shapes, normalization, flattening, and visualization, then demonstrates why raw pixel similarity can mislead.
- Examples
- Uses MNIST handwritten digits, 28 × 28 pixel arrays, heat-map visualizations, and three contrasting digit examples.
- What you can do after
- Apply it via PyTorch and torchvision to load, normalize, reshape, visualize, and prepare image tensors for models.
Read how Bayesian Optimization in Action explains digital image →
Outlier Detection in Python
Image data is visual data represented as arrays of pixel values or derived visual features.
- How it is used
- Identifies novel objects outside the training distribution and moderates confidence in image-based predictions.
- How deep it goes
- Distinguishes object novelty from unusual appearance and outlines five detection approaches without detailing their implementations.
- Examples
- Uses autonomous-driving objects, bird classifiers, and rotated, discolored, cropped, or stretched images as detection examples.
- What you can do after
- Apply image outlier detection through reconstruction error, model confidence, embedding detectors, generative models, or perturbation tests.
Read how Outlier Detection in Python explains digital image →
Inside Deep Learning: Math, Algorithms, Models
Data whose values are arranged to represent images, typically as spatial grids of pixels or channels.
- How it is used
- Organizes spatial data so convolutional neural networks can use image structure for classification.
- How deep it goes
- Defines channel and spatial dimensions and connects them to convolution filters, stopping short of a complete tensor implementation.
- Examples
- Uses color images, line and grid patterns, and a two-channel sliding-filter example.
- What you can do after
- Understand how to organize image channels and spatial dimensions for convolution-based image classification.
Read how Inside Deep Learning: Math, Algorithms, Models explains digital image →
AI-Powered Search
An image is visual data represented as a grid of colors that forms a picture.
- How it is used
- Uses images as searchable unstructured data and as inputs for multimodal search.
- How deep it goes
- Defines images as color grids and outlines label-based and vector-based retrieval, stopping short of embedding-model training details.
- Examples
- Uses color grids, image labels, text descriptions, stored images, and query images as concrete artifacts.
- What you can do after
- Understand image retrieval through labels or embedding vectors and distinguish visual similarity search from text-based indexing.
Data Storytelling with Altair and AI
An image is a visual representation stored or rendered as graphical content.
- How it is used
- Uses images to clarify a data story’s subject, reinforce context, and guide audience emotion.
- How deep it goes
- Explains visual purpose, copyright concerns, generation, selection, and chart insertion; defers advanced generation techniques.
- Examples
- Uses generated homelessness photos, an illustrated girl in a meadow, and sports images embedded in a bar chart.
- What you can do after
- Apply it via DALL-E and Altair to generate, select, host, and place visuals in data-story charts.
Read how Data Storytelling with Altair and AI explains digital image →
Web Design Playground, Second Edition
A visual representation embedded in or associated with a web document.
- How it is used
- Adds visual content to web pages through separately stored files referenced by HTML.
- How deep it goes
- Explains img attributes, accessibility text, responsive sizing, and photo preparation without detailing image formats internally.
- Examples
- Uses a branded logo, fluid responsive images, and gallery thumbnails paired with full-size photographs.
- What you can do after
- Build web pages with accessible img elements and prepare images for responsive layouts and photo galleries.
Read how Web Design Playground, Second Edition explains digital image →
Deep Learning with R, Third Edition
An image is visual data represented as a structured arrangement of pixel values.
- How it is used
- Represents collections of images in a tensor-oriented data structure.
- How deep it goes
- Defines the rank-4 shape and pixel-channel organization without explaining channel meanings or image-processing operations.
- Examples
- Uses a rank-4 image-data tensor diagram as its concrete representation.
- What you can do after
- Understand how image datasets map onto sample, height, width, and channel axes.
Read how Deep Learning with R, Third Edition explains digital image →
Grokking AI Algorithms, Second Edition
Image data is data that represents visual content through digital image values and associated metadata.
- How it is used
- Supports training generative image models with visual examples and associated captions.
- How deep it goes
- Defines numerical representation and surveys collection sources, but stops short of explaining pixel encoding or model internals.
- Examples
- Uses paintings, landscape photos, museum scans, medical scans, satellite imagery, 3D renders, and 8×8 grayscale faces.
- What you can do after
- Evaluate image datasets for scale, diversity, caption accuracy, and relevance to intended image generation.
Read how Grokking AI Algorithms, Second Edition explains digital image →
Writing for Developers: Blogs that get read
A two-dimensional visual representation presented to communicate information in digital or physical form.
- How it is used
- Uses images to explain engineering systems, data, optimizations, debugging clues, and implementation decisions.
- How deep it goes
- Covers selection, cropping, captions, color accessibility, clarity, sizing, and ALT text without teaching data visualization design in depth.
- Examples
- Uses architectural diagrams, optimization graphs, dashboard screenshots, charts, and implementation visuals as concrete examples.
- What you can do after
- Evaluate images by checking reader focus, readability, captions, color accessibility, clarity, size, and ALT text.
Read how Writing for Developers: Blogs that get read explains digital image →
How Large Language Models Work
An image is a structured visual representation composed of values that encode colors or other pixel properties.
- How it is used
- Represents images as vector sequences for transformer-based computer vision processing.
- How deep it goes
- Explains pixels, RGB values, fixed-size patches, vector dimensions, preprocessing, and reconstruction.
- Examples
- Uses 16 × 16 patches and calculates their resulting 768-value RGB vectors.
- What you can do after
- Understand how images become numeric patch sequences suitable for transformer processing.
Read how How Large Language Models Work explains digital image →
Related concepts
- DALL-Eproduced by · DALL-E is a family of OpenAI generative models that creates and edits images from natural-language descriptions and other conditioning inputs.
- linear gradientincludes · A linear gradient is a CSS-generated image whose colors transition along a straight line between specified color stops.
- pixelhas part · A pixel is either a discrete sample carrying visual information at a position in a digital image or a reference unit used to express rendered dimensions.
- captionused by · A caption is text associated with visual or audiovisual content that identifies, describes, or represents its content.
- multimodalused by · A multimodal model is a model that processes, generates, or relates information represented in two or more data modalities, such as text, images, audio, or video.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.