yuhuan-wu / rdpnet Goto Github PK

When we will get the code of this paper?

When I run ‘python setup.py build develop', I got this error'RuntimeError: Error compiling objects for extension'
my env ：pytorch==1.7.1 torchvision==0.8.2 torchaudio==0.7.2 cudatoolkit=10.2
but when I use your env'pytorch==1.1 torchvision==0.2.2 cudatoolkit=9.0',I successfully run the build file.However, when I run the 'demo/visualize.py'file, I got another error'undefined symbol: _ZN6caffe26detail36_typeMetaDataInstance_preallocated_7E',because the CUDA environment of my system is 10.2.
So I want to ask you, can this code run only when CUDA is 9.0？

Hi, I'm very appreciate of your impressive work! However, I just found something inconsistent about the training setting of iSOD dataset.

In section 4.A of your paper, you said that you use 500 images for training and 200 images for validation. But the JSON file of iSOD of the provided dataset contains 700 images for training. I wonder if the training JSON is not the one used in your paper? Thank you very much.

Hello!I can't get the data from Google ,so i want to ask you for help,but your e-mail is wrong!Can you tell me the right e-mail?Thank you very much!

Compared to https://github.com/facebookresearch/maskrcnn-benchmark, this folder is missing.
After compiling, the code cannot run correctly.
It will occur 'ImportError: cannot import name 'datasets'.
Is it caused by the missing files?