1 paper
Volodymyr Fedynyak, Yaroslav Romanus, Bohdan Hlovatskyi +4
The recent works on Video Object Segmentation achieved remarkable results by matching dense semantic and instance-level features between the current and previous frames for long-ti…