Hi,
we are planning to build a tracking system for a digital fitness workout app. In assessment of computer visionn vs. sensor solutions, we would need to know the limitations of fitness (bodyweight / strength / cardio) exercises that can be recognized and measured (repetitions, speed, ...).
As most sensor motion tracking/ capture systems require multiple sensors to capture skeletal movement + posture, we are wondering about the limitations of the BHI260AP self-learning AI sensor, as this would only be attached on one body part. So we would guess that even with the ability to learn new exercises, some exercises could not be tracked depending of the location of the sensor? Do you have any further specifications of that, e.g.
- Where (body point) would the sensor need to be worn to be able to recognize a maximum set of fitness exercises?
- What exercisescan normally not be tracked (e.g. left arm exercise if sensor is on right arm)
- How accurate is the tracking (time lag, % of repetitions recognized, ...)?
Could you give any specifications (besides the 15 example workouts) of workouts that are already possible with the sensor or can be learned by it?
We would love to use the sensor as it would be a very convenient setup for the user, however need to ensure a good tracking accuracy in terms of depth and breadth, thus thank you for your answers.
If more convenient, happy to agree to a call.
