System Technical Specs
The Yakkertech ball tracker uses a proprietary stereo vision imaging camera system to capture images of the ball as it flies past, then determines the position of the ball in 3D and analyzes the seams on the ball to determine its rotation. By capturing the position and rotation, the device is able to determine the velocity, spin rate, spin axis and trajectory of the ball, and compute the ball trajectory. This information is then reported to the user in TrackMan and PITCHf/x compatible data formats.
The device is precisely calibrated by measuring a calibration target that is placed on the field. It sits on a tripod during use, and its optical axis is perpendicular to the line of flight of the ball.
The accuracy of the Yakkertech ball tracker has been validated using computer-generated pitch information and photo-realistic images of the ball. A trajectory model was used to calculate the position of a baseball in time for twenty pitches with randomly-generated initial velocity, release position, strike-zone position, spin rate, and spin axis. Using the positions and velocities from the calculated ball flight, the exact kinematics of the ball were determined mathematically. The positions and rotations of the ball were then put into 3D rendering software to generate photo-realistic images of the ball, as seen by a computer model of the ball tracker’s cameras. This model included lighting to simulate the hard shadows seen in bright daylight and a specially-extracted real baseball texture to accurately reproduce the seams. A stereo calibration of the modeled ball tracker was performed using a real texture of the calibration board using the same technique we use for the real stereo calibration. Similarly, a field calibration of the modeled ball tracker was performed using the same calibration technique that is used with the real system.
The computer-generated images of the calibration procedures and ball pitches were then fed into the ball tracker software as if the ball tracker’s cameras had captured them directly. The software analyzed the pitch images and produced ball kinematic data which was sent to the front-end software. We then compared the kinematic data reported by the mathematical analysis of the ball trajectory (original data) with the kinematics reported by the ball tracker software’s analysis of the generated images (measured data). This allowed us to analyze the accuracy of the ball tracker with respect to ground-truth data, as the precise kinematics of the computer-generated data and ball images were known. In addition to synthetic validation, the system has been compared to the data produced by a TrackMan system and produces very consistent results.
Proprietary table of validation metrics is available upon request.
- Interface computer requirements: Android Device
- Data formats: TrackMan, PITCHf/x
- Pitch computation time: about 3 seconds
- Cameras: Basler AC1440’s
- Frame rate: 220 fps per camera
- Resolution: 1440×1080 per camera
- CPU: 6th generation Intel Core i5 quad core at 3.3 GHz
- Memory: 8 GB
- SDD: 256 GB
- Weight: 9 lbs
- Dimensions: 24″ long, 4″ high, 8″ deep
- Connectivity: 802.11 WIFI with WPA2 encryption, wired ethernet
- Power supply: 120 volts, 100 watts