Dahun Kim comments

Results 15 comments of


                                            Dahun Kim

[DITO] pre-computed embeddings

Hi, the demo with custom text embeddings is now available at [here](https://github.com/google-research/google-research/blob/eae296c7e3668db1fceab3c0e371b1714f87a286/fvlm/dito/demo.py#L98)

Pretrained Model Setting

Hi, I will soon upload the image "Base" model definition and configuration for reference. Thanks.

Hi, The VIPER dataset was first published at ICCV 2017. Please refer to their website below. https://vladlen.info/publications/playing-for-benchmarks/ We post-process with their instance segmentation and semantic segmentation datasets. The plan for...

VIPE dataset

Sorry for the delay. The converted dataset is sent via email.

VIPE dataset

Please request via [email protected], so that I can know your email address, thanks.

Cannot run code!

@[sagar1garg ](https://github.com/sagar1garg) Please check the latest update with Pytorch1.4 and Cuda10.0. Thanks.

could you please tell me version of cuda and cudnn？

@[sagar1garg ](https://github.com/JiangYi0311) Please check the latest update with Pytorch1.4 and Cuda10.0. Thanks.

mmdet-version question

Hi, The mmdetection used in this repo is `Tag: v1.0rc1` which requires mmcv > 0.2.10. However, I find mmcv >= 0.2.15 raises compile errors, so I recommand mmcv==0.2.14.

Demo script for inference on your own video

Thanks. Let me get back to this in a few weeks.

Demo script for inference on your own video

> I wanted to run this against this video https://archive.org/details/0002201705192 and share results on my YT channel. > After 1h of fighting with code I know that I will not...