Can you provide a pretrained model for test model?

I've read your paper, it's a fantastic job. But at present, the stability of the algorithm cannot be determined only by looking at the image results. I would like to do some tests in video scenarios based on your model，can you provide a pre-trained model?