Search for a command to run...
Modeling Motion with Multi-Modal Features for Text-Based Video Segmentation