WeVisOne-VL: visual perception from boxes to alpha mattes
Segment objects in videos using text descriptions