qinxin
qinxin
Hi, thanks for your awesome work. Could you describe your ensemble strategies more detailly? I also tried some strategies to ensamble different models, but the results are far away from...
In the action localization task，you provided the code to get the AUC metric for action proposal evaluation. Could you also provide the classification results to get the mAP?
> As our recent researches use this code to detect human face landmarks, Could you please provide the citation of this code? 朋友你好， 我关注了你们的论文FaceSwapNet， 我想请问一下你们的关键点是怎么得到的，这里不是没提供win版本吗？
In my experiments in video classification，non local moudle is not query-independent. How about you guys' results?
This error is raised because the pickle version. It's abvious that the file path you read is encoded by byte mode by the prefix 'b'. You should convert it to...
These files are so big so you should download them by yourself as I writed in READEME.
Please read the README carefully. I already offered the link to download the CSV files.
Yes, you are right. You need another module to classify the proposals which is not implemented in my code
> > > Yes, you are right. You need another module to classify the proposals which is not implemented in my code > > > > > > could you...
"PCM模块中先使用了一个Attention模块来提取上下文的语义信息，然后同时又使用了一个Attention模块来提取位置敏感的信息，这样的话一个PCM中包含两个Attention", 语义上下文和位置上下文是在同一个模块中实现的，batch_size=32, 刚好占满两张1080Ti的显存。 我是有开源的计划的，但是paper一直没中，所以迟迟没有开源，paper中之后或者我毕业之后会开源。