yunbo426/mim

Code release for "Memory In Memory: A Predictive Neural Network for Learning Higher-Order Non-Stationarity from Spatiotemporal Dynamics" (CVPR 2019)

mim's Issues

Can you provide Radar Echo Pre-trained model?

Commands to train?

Hi, nice paper. For the github, could you please provide the training commands that you used?

About the taxibj dataset?

Hello, I want to know that the channel of taxibj dataset is 2. How to calculate the ssim between the predict frames and input frames?

could you provide the full experiment settings for all the datasets?

Interesting work! Could you please provide the full hyper-parameters setting for TaxiBJ and other datasets in your experiment? I want to reproduce your results but I am not very clear when I read the papers.

Will you release the training code?

Can you provide Radar Echo Dataset?

Hi, I saw Radar Echo Dataset in the paper.
Where can I download the Radar Echo Dataset? Thanks.

raise ValueError('Name of network unknown %s' % name) ValueError: Name of network unknown convlstm_net

The program can only run when the ’model_name‘(in run.py line44) is ’mim‘.
The program cannot run when the name is other, such as [convlstm, predcnn, predrnn, predrnn_pp, convlstm_net](in run.py line 43).
So I think we should change 'convlstm_net' to 'mim'(in run.py line44).
Could you share the code of other models[convlstm, predcnn, predrnn, predrnn_pp, convlstm_net]?

求教几个小问题

大佬些,我看到论文中预测的帧存在结果比较模糊的问题,请问是否考虑过类似RNN+GAN的模型来预测生成更清晰的未来帧?

关于基线数据集,除了移动MNIST,以及低像素的Human3.6M,是否尝试过更高分辨率的图像?

另,README.md中的图片挂了.

self.params.update(self.configs.__dict__['__flags']) KeyError: '__flags'

self.params.update(self.configs.__dict__['__flags'])

KeyError: '__flags'

Visualise results

Hi,
Thank you for the work. I have a question about TaxiBJ dataset. How do you visualise results? When I save the prediction, it's grayscale.
Please advise.

Thank you!