Jonas Wu

icon indicating an email [email protected]

icon company The University of Hong Kong icon location China icon location The University of Hong Kong. PhD student. Computer Vision. Multimodal Models.

Results 5 issues of Jonas Wu

About distributed training

1 comment

Hi, is this project able to use distributed training in multiple nodes?

datasets for the first-stage autoencoders

1 comment

Thanks for the wonderful work! I want to to know what datasets are used to train the first-stage autoencoders, as shown in https://github.com/CompVis/latent-diffusion/blob/main/scripts/download_first_stages.sh. Thanks.

About randomly sample support images during inference

Hi, thanks for your clear codebase. I have a question regarding the inference. It seems that the dataloader randomly samples a support image, e.g. [here](https://github.com/Seokju-Cho/Volumetric-Aggregation-Transformer/blob/f54c9df3da6d3b261ba5091460ab9f6cc168c80b/data/coco.py#L88), which could lead to different...

Question about test

3 comment

Hi, I ran the code of test, and the results are shown as blow. ![image](https://user-images.githubusercontent.com/21001460/132433338-27430902-fae8-4812-b23e-b9f09c7233d5.png) I wonder whether the repo supports evaluation of mAP of iou@[03:0.7], as reported in the...

Features link can not open

2 comment

Hi, Thanks for your code and features. But the feature links (THUMOS [https://jbox.sjtu.edu.cn/l/pn3mvh](https://jbox.sjtu.edu.cn/l/pn3mvh) and Anet [https://jbox.sjtu.edu.cn/l/vuB3WW](https://jbox.sjtu.edu.cn/l/vuB3WW)) provided in your README can not open. Could you please provide another link? And...