What is exercises-dataset?

Exercises Dataset is a ready made collection of information about 1,324 fitness exercises, meant for anyone building a workout app, doing fitness research, or working on a machine learning project related to exercise. Each exercise comes with a short animation GIF, a small thumbnail image, the body part and muscle it targets, the equipment needed, and step by step instructions written out in ten different languages including English, Spanish, Chinese, and Hindi. The data was originally built to power a separate mobile app called LogPress, an AI assisted workout tracker, but the README explains it can be dropped straight into anyone else's own backend. The main file is a single JSON document listing every exercise as a structured record, along with a JSON Schema file that describes exactly what fields each record contains so the data can be validated automatically. The repository also includes two browser based tools that need no server or installation to use. One is a searchable, filterable exercise browser you simply open in a web browser, letting you look through every exercise by category, equipment, or targeted muscle. The other is a developer setup guide that generates ready to run database import scripts for several common databases, produces sample client code in multiple programming languages for calling a backend API, and even provides a prompt you can paste into an AI assistant like ChatGPT or Claude to have it generate a working backend API for this dataset. According to the included statistics, exercises are spread across body parts such as upper arms, upper legs, and back, and roughly a quarter of them require no equipment at all, making the dataset useful for apps focused on home workouts. The code and data are released under the MIT license, while the animation GIFs and thumbnail images are used under separate terms from their original creator, Gym visual, as explained in the repository's notice file.