How can I improve the output with my custom video??

[Samuel3Shin]

I get some bad output from my video.

I think the model makes incorrect panoramic images, therefore the final results are not good as well...

The panoramic_robust_mean.png are like below.

Is there some way to improve outputs from my custom videos by manipulating the hyperparameter or something??

Thanks!!!

P.S
Those are my run command, so I used all the default arguments.
python src/DeepMCBM.py --dir video1
python src/DeepMCBM.py --dir video2

[guy-erez]

You should consider reducing the learning rate. It is highly recommended to track the alignment status using some asynchronous logging mechanism. A version that supports Neptune.ai may be released in the future.