implementation of video inpainting with hybrid approach

I want to implement removal of object using video inpainting with hybrid approach like optical flow and convolution neural network. Input video should be with occlusion. Need proper algorithmic steps. and results should be in terms of parameter. so can easily compare with existing one.