1 paper · 1 filter
Yusuke Oumi, Yuto Shibata, Go Irie +3
Can we recover the 3D poses of multiple people using only sound? This paper presents the first attempt to estimate multi-person 3D poses solely from acoustic signals. Estimating th…