The official implementation of "LTCA: Long-range Temporal Context Attention for Referring Video Object Segmentation".