hdidwania/mfsGAN

Multi-frame and multi-scale conditional generative adversarial networks for foreground extraction

Multi-frame and Multi-Scale Conditional GAN for foreground extraction

Code for multi frame multi scale conditional GAN used in the papers Multi-frame and Multi-scale Conditional Generative Adversarial Networks for Efficient Foreground Extraction and GAN based efficient foreground extraction and HGWOSA based optimization for video synopsis generation for foreground extraction.

The core requires data in format of CDNet dataset.