首页|'Training One or More Machine Learning Models to Recognize One or More Movements Using Virtual Actors and Virtual Cameras' in Patent Application Approval Proces s (USPTO 20240233443)
'Training One or More Machine Learning Models to Recognize One or More Movements Using Virtual Actors and Virtual Cameras' in Patent Application Approval Proces s (USPTO 20240233443)
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
The following quote was obtained by the news editors from the background informa tion supplied bythe inventors: “In order to train machine learning models to co rrectly identify people in specific bodypositions, example training data is nee ded that expresses a range of variability for how the position maybe viewed in a live application. Traditionally, training data for body position would be capt ured by editingselections of live actors in example poses. In real-world applic ations, users will be found in a variety oforientations. For example, the user may not be centrally positioned in a frame, some body parts may be outof frame, the user’s body may be rotated or not directly facing the camera, etc. Training a machine learningmodel to recognize these variance requires training data for each scenario. As an example, capturing thesevariances would require multiple videos and/or images accounting for each variance. Capturing imagesand/or video s of live actors in these situations, and labelling the data correctly, is both costly and timeconsuming. Moreover, some variances may be missed, which may cau se the machine learning model tonot recognize and/or misidentify the body posit ion. Furthermore, existing machine learning models arelimited to detecting sing le movements and cannot recognize or detect movements or compound workouts,such as superset workouts where an athlete alternates sets of at least two different exercises.”
NETL
NSTL
