yangshao

@B2Gdevs This solution seems weird to me? so you assume the model has to predict a bounding box close to a given bounding box? what if the model cannot... For...

@jihopark do you solve this problem?

yes，i checked it. besides， do you know howto do beam search in tf 1.1.0？ thanks On Jul 14, 2017 11:05 PM, "Jaemin Cho" wrote: > Did you check argument when...

hi, did you solve this problem, I think I have the same problem. but no idea.

In addtion, after doing the step: "Preprocess the dataset using tools/preprare_data.py" to generate files such as reltest.json seems useless to run code like test_predicate_recognition.py

Thanks, can you give more details on what are the differences between test.pkl and proposal.pkl?

and If proposal.pkl is generated using faster rcnn, how do you decide the number of bbox in the test samples?

Hi, I have more questions: 1. what's the representation of bbox: (x1, y1, x2, y2) or (x1, y1, width, height)? 2. how to use word embedding or what's the format...

I find my problem: I forget to add for the decoder input! currently it works well. Can you explain what's the meaning of logits.sample_id? why not use rnn_output? Thanks