An IMU dataset of athletic hammer throwing with phase-segmented kinematic events
This paper presents a public dataset comprising inertial measurement unit (IMU) recordings from 11 Spanish hammer throwers, featuring four-sensor data at 120 Hz and manually verified timestamps for nine key biomechanical events to support research in sports biomechanics and human movement analysis.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer
In the world of sports science, understanding how the human body moves is often a matter of seeing the invisible. For decades, coaches and researchers have relied on high-speed cameras and specialized laboratories to break down complex athletic feats into measurable parts. This approach, known as biomechanics, allows experts to see exactly when a foot leaves the ground or how a joint rotates, but it usually requires the athlete to perform in a controlled room filled with equipment, far from the reality of a training field. While this data is precise, it is difficult to gather during regular practice, limiting how much it can help athletes improve their daily routines. The challenge has been finding a way to capture this same level of detail using tools that are small, affordable, and portable enough to wear on the body without getting in the way.
A team of researchers from Spain has now addressed this challenge by creating a new public record of data from one of the most dynamic events in track and field: the hammer throw. In this event, an athlete spins rapidly inside a circle, building momentum with a heavy metal ball attached to a wire, before releasing it as far as possible. The technique is a delicate sequence of turns, where the thrower shifts their weight between standing on one foot and standing on both, all while accelerating the heavy implement. To capture this motion, the researchers attached small, lightweight sensors to the bodies of eleven elite Spanish throwers. These sensors, known as inertial measurement units, act like tiny digital observers that record acceleration, rotation, and direction hundreds of times every second. By placing these devices on the wrists, feet, and lower back, the team was able to map the entire throwing motion without needing a camera or a laboratory.
The result is a detailed dataset that breaks down every single throw into nine specific moments, from the very first shift of weight to the final release of the hammer. The researchers recorded these movements from eleven athletes, including seven men and four women, who competed at regional and national levels. Each athlete performed throws using standard competition equipment, with the weight of the hammer adjusted to match their age and gender, ranging from three kilograms for some women up to 7.26 kilograms for the men. The sensors were placed on the right wrist, the right and left feet, and the lower back, with an optional fifth sensor sometimes attached to the hammer wire itself. As the athletes spun, the sensors captured the raw data of their motion, which was then processed to identify exactly when the feet left the ground and when the hammer was let go.
To ensure the data was accurate, the team did not rely on the sensors alone. They cross-checked the computer-generated timing of each movement against the actual video recordings of the throws. In nearly every case, the automated detection matched what the researchers saw with their own eyes. When the sensors detected a moment where the thrower was standing on one foot versus two, the visual evidence confirmed it. The only time the automated system needed a human hand to correct the timing was in a handful of instances where the signal was slightly confusing, such as when the hammer was released. Even then, the researchers could look at the video to fix the timestamp, ensuring that the final record was precise to within a fraction of a second. This careful verification process means that the dataset is not just a collection of numbers, but a reliable map of how elite athletes actually move.
What makes this work particularly valuable is that it is the first time such a detailed record of hammer throwing has been made available to the public. Before this, researchers studying this sport had to build their own sensors or rely on expensive, bulky equipment that was hard to use outside a lab. Now, anyone interested in sports science, computer programming, or athletic training can download these files and study the motion of the hammer throw. The data includes the raw signals from the sensors as well as a separate file that lists the exact timing of the nine key events for each throw. This allows scientists to test new ways of analyzing movement, train computers to recognize athletic techniques, or simply compare how different athletes approach the same physical challenge.
The researchers also checked to see if their findings matched what was already known about the sport. They compared the timing of the turns in their dataset with data from world-class finalists who had been filmed in previous championships. They found that the pattern of movement in their group of athletes was consistent with the best in the world, even though the athletes in this study were not all world champions. The throwers in the study spent a slightly longer portion of their turns standing on both feet compared to the elite finalists, a difference that makes sense for athletes who are still developing their technique. This comparison confirmed that the sensors were capturing the true structure of the throw, validating the method for future use.
By making this data open to everyone, the team has removed a significant barrier to research in sports biomechanics. The files are stored in a standard format that can be read by common computer programs, and the sensors used are inexpensive enough that other researchers can replicate the setup. The dataset covers a wide range of athletes and conditions, providing a solid foundation for understanding the physics of the hammer throw. It offers a clear view of how the body coordinates a complex, high-speed rotation, turning a fleeting moment of athletic power into a permanent record that can be studied, shared, and improved upon. This work does not just describe a single event; it provides the tools for the next generation of scientists to understand human movement in a way that was previously impossible outside of a specialized laboratory.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.