search

snap-research / NeROIC

Other
960 stars 124 forks source link

Results not as good on custom dataset #23

Open anna-debug opened 2 years ago

anna-debug commented 2 years ago

Hello! I have been training the network on my own data and the rendering results are way worse than the ones from training on the provided datasets. I trained on a dataset that has 53 images. I kept the same parameters for the config files, changing the image size and name, of course.

One of the input images: 2 And these are the result images taken from the logs folder: 00 train_01 Visualizing the dense output from colmap: Screenshot from 2022-09-08 11-26-43

What could be the possible reasons for such results?

zfkuang commented 2 years ago

Hey anna,

I can't tell what's going on based on the images you're sending me. Can you send me your log files? (it would be better if you could also send your training images, you can send them to my email: zhengfei@stanford.edu)

Also, please make sure you are running your code under ubuntu. Windows is not supported for now.

pfirmino commented 1 year ago

Any plans to have a working windows release?