Algorithm 1: 3D Yolo-KF-Tracking

Notation: object states $x_k$, measurement $z_k$, Kalman filter KF, image set $F$ Input: image $F$ while true do   Object-Detection (F)   if object detected then     trigger and initiate KF     break   else     continue   end if end while while true do   KF.predict()   Object-Detection (F)   if object detected then     if confidence score > 0.75 then       ${\widehat{x}}_k$ = $z_k$       KF.update ($z_k$)     else       KF.update ($z_k$)       ${\widehat{x}}_k$ = KF.update ($z_k$)     end if   else     ${\widehat{x}}_k$ = KF.update (KF.predict())   end if   Output: ${\widehat{x}}_k$ (posteriori estimate)   continue end while