STRENGHT TRAINING WITH RESISTANCE
Initially, these studies were not considered, but due to the delay caused by the need to carry out new hardware development, commercial inertial sensors were used, which, despite their low sampling frequency, allow for the monitoring of strength training with loads. The specific objective was set as follows:
Recognition of exercise type through artificial intelligence and smartwatches.
This objective has been successfully achieved, and the results will be presented at the international conference specialized in artificial intelligence. In this work, the Amazon S.A. smartwatch Fitbit Versa 3 (Figure A) was used, which was reprogrammed to access its accelerometer data.
By training an artificial intelligence engine through deep learning, a classification accuracy of over 96% was achieved for various gym strength training exercises, distinguishing between biceps, triceps, chest, and back exercises, as shown in the concordance matrix in Figure B.
(A) Smartwatch Fitbit Versa 3. (B) Confusion Matrix Obtained.
