Yuren Cong comments

Results 46 comments of


                                            Yuren Cong

ImportError: cannot import name '_C' from 'fasterRCNN.lib.model' (/home/k/kajal/STTran/fasterRCNN/lib/model/init.py)

I guess you didn't compile the C++ code (correctly). Please run the setup.py. Btw, I think the code only supports the cuda version

ImportError: cannot import name '_C' from 'fasterRCNN.lib.model' (/home/k/kajal/STTran/fasterRCNN/lib/model/init.py)

> Do you solve this problem? My cuda version is 11.8 and I also met this problem. I think the exclusive solution is using cuda

ImportError: cannot import name '_C' from 'fasterRCNN.lib.model' (/home/k/kajal/STTran/fasterRCNN/lib/model/init.py)

Have you compiled the code again after downgrading the cuda version?

the dataset [Action Genome](https://www.actiongenome.org/#download)-- this link is unavailable

I have no idea... Please contact the authors of Action Genome. Or you can have a try here: https://github.com/JingweiJ/ActionGenome

save graph and graph visualisation

Do you mean to save the graphs as images? You could implement it with GraphViz

save graph and graph visualisation

You can directly convert the tensors to numpy arrays and save them.

What happens when there are no relations in a sample?

You can easily add a dummy predicate label "no_relation" and combine random bboxes with this dummy label.

请问如何将inference.py得到的场景图保存成一个json文件呢

您可以把triplets保存成json，如果是图结构的话要结合GraphViz

RuntimeError: CUDA error: device-side assert triggered CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect. For debugging consider passing CUDA_LAUNCH_BLOCKING=1. Compile with `TORCH_USE_CUDA_DSA` to enable device-side assertions.

Hi, If you haven't changed any of the code, you might be using an incompatible version of cuda. Please make sure your cuda version matches your GPU device. For example,...

RuntimeError: CUDA error: device-side assert triggered CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect. For debugging consider passing CUDA_LAUNCH_BLOCKING=1. Compile with `TORCH_USE_CUDA_DSA` to enable device-side assertions.

> 同样的问题，coco的数据加载会过滤左上角大于等于右下角的box，导致一些box被删除了。但是relation里面box的索引还在，就导致了cuda的越界警告。使用作者目前给的标注会有这个问题。 wow，我不知道这个情况，是由于pycocotools代码改变引起的吗？