A comprehensive fitness exercise dataset containing 1,324 exercises, each with an animation GIF, 180×180 thumbnail, and step-by-step instructions in nine languages, has been released on GitHub. The dataset powers the LogPress AI-assisted workout tracker and is available for developers to integrate into their own fitness applications.
Dataset Contents and Structure
The dataset includes 1,324 exercise records with metadata such as category, body part, equipment type, target muscle, and secondary muscles. Each entry comes with a local 180×180 thumbnail image and an animation GIF, both sourced from Gym Visual and used with permission. Instructions are provided in English, Spanish, Italian, Turkish, Russian, Chinese, Hindi, Polish, and Korean. The data is stored as a JSON array in `data/exercises.json`, accompanied by a JSON Schema (Draft 2020-12) for validation. Approximately 25% of the exercises require no equipment, making the dataset suitable for at-home workout applications.
Developer Tools and Integration
The repository includes two ready-to-use HTML files that require no server. `index.html` is an interactive exercise browser with live search, filtering by category, equipment, and target muscle, and an infinite scroll grid. Clicking any exercise card displays full details and instructions in any of the nine languages. `setup.html` provides a developer guide with database setup scripts for SQL Server, PostgreSQL, MySQL, and SQLite, generating a ready-to-run SQL file with all 1,324 INSERT statements. It also offers API integration code snippets in JavaScript, Python, C#, Java, PHP, Go, and cURL, and a structured prompt for generating a production-ready REST API using frameworks like Express.js, FastAPI, ASP.NET Core, Spring Boot, Laravel, or Gin via LLMs such as ChatGPT, Claude, or Gemini.
Sample Exercises and Usage
The dataset covers a wide range of exercises, from the 3/4 sit-up (targeting abs with body weight) to the barbell bench press (targeting pectorals), barbell deadlift (targeting glutes), and barbell full squat (targeting glutes and quadriceps). Each sample includes key cues and secondary muscle activation. The dataset is intended for building fitness or workout planning applications, machine learning projects for exercise recognition or recommendation, health and wellness research, and educational demonstrations.
Licensing and Attribution
The code and data are released under the MIT license, while the media files (images and GIFs) are © Gym Visual and used with permission, as detailed in the NOTICE.md file. The repository is structured with a `data/` folder for JSON files, `images/` and `videos/` folders for media assets, and the two HTML tools in the root directory.