3 ms·
I'm going to guess they start with a generic human model that includes all limbs and extremities and then the "machine learning" process attempts to fit that mo
by RodgerTheGreat 9y ago
I'm going to guess they start with a generic human model that includes all limbs and extremities and then the "machine learning" process attempts to fit that model to the silhouettes extracted from the video.
- stochastic_monk 9y agoWhich implies that the technique uses domain knowledge of people to make assumptions about their morphology.