Semi-supervised video object segmentation evaluation in python

MiVOS (CVPR 2021) – Mask Propagation

[CVPR 2021] MiVOS – Mask Propagation module. Reproduced STM (and better) with training code . Semi-supervised video object segmentation evaluation.

Ho Kei Cheng, Yu-Wing Tai, Chi-Keung Tang

New! See our new STCN for a better and faster algorithm.

ParkourBike

This repo implements an improved version of the Space-Time Memory Network (STM) and is part of the accompanying code of Modular Interactive Video Object Segmentation: Interaction-to-Mask, Propagation and Difference-Aware Fusion (MiVOS). It can be used as:

  1. A tool for propagating masks across video frames. Results
  2. An integral component for reproducing and/or improving the performance in MiVOS.
  3. A tool that can compute dense correspondences between two frames. Tutorial

Overall structure and capabilities

MiVOS

 

To finish reading, please visit source site