@chaofeng_cs Profile picture

Chaofeng

@chaofeng_cs

Postdoc @NTUsg, PhD @HKUniversity

Similar User
Yuekun Dai photo

@YuekunDai

Ziqi Huang photo

@ziqi_huang_

Yuanhan (John) Zhang photo

@zhang_yuanhan

Li Bo photo

@BoLi68567011

Davide Moltisanti photo

@davmoltisanti

Jiaqi Wang photo

@wjqdev

Xiangtai Li photo

@xtl994

Lingdong Kong photo

@ldkong1205

Jiahao Xie photo

@JiahaoXie3

Guangcong Wang photo

@GuangcongW

Xintao Wang photo

@xinntao

Fangzhou Hong photo

@hongfz16

Yuhang Zang photo

@yuhangzang

Kelvin Chan photo

@kelvinckchan

Haoning Wu photo

@HaoningTimothy

Chaofeng Reposted

github.com/Q-Future/A-Ben… “Kindly blind at weird generations?” “Sadly yes.” We demonstrated that LMMs, though widely considered as better T2I evaluators, are in-turn especially less sensitive on common types of T2I-generated errors. Arxiv: arxiv.org/pdf/2406.03070

Tweet Image 1

Chaofeng Reposted

Our @gradio demo for Q-Align is embedded in our homepage! Visit Q-Align.github.io! It can: - Rate a score for an image or a video. - Predict the probabilities of rating levels!

Tweet Image 1

Chaofeng Reposted

Q-Instruct + TexForce + SD-Turbo. Making better quality AIGC (latter for each pair) faster.

Tweet Image 1
Tweet Image 2
Tweet Image 3
Tweet Image 4

Improve sd-turbo in one line: PeftModel.from_pretrained(pipe.text_encoder, 'chaofengc/sd-turbo_texforce') Arxiv: arxiv.org/abs/2311.15657 Github: github.com/chaofengc/TexF… huggingface: huggingface.co/chaofengc/sd-t…

Tweet Image 1

Chaofeng Reposted

The Visual Scorer API from Q-Instruct: github.com/Q-Future/Q-Ins…

Tweet Image 1

The HF space is also updated (corresponding to the newer version): huggingface.co/spaces/teowu/Q… (still the link)



Chaofeng Reposted

The HF space is also updated (corresponding to the newer version): huggingface.co/spaces/teowu/Q… (still the link)

github.com/Q-Future/Q-Ins… We release a better version of weights (totally reproducible) for Q-Instruct!



Chaofeng Reposted

Thanks @_akhaliq for the sharing. Check out our MVHumanNet, the largest-to-date dataset of multi-view human captures, with 4,500 human identities and 9,000 daily dressing. We plan to release it in the next months. Following our MVImgNet, hope it can help.

MVHumanNet: A Large-scale Dataset of Multi-view Daily Dressing Human Captures paper page: huggingface.co/papers/2312.02… In this era, the success of large language models and text-to-image models can be attributed to the driving force of large-scale datasets. However, in the realm of…

Tweet Image 1


Thank you @_akhaliq for sharing our work😄!

Q-Instruct: Improving Low-level Visual Abilities for Multi-modality Foundation Models paper page: huggingface.co/papers/2311.06… Multi-modality foundation models, as represented by GPT-4V, have brought a new paradigm for low-level visual perception and understanding tasks, that can…

Tweet Image 1


Chaofeng Reposted

Q-Instruct: Improving Low-level Visual Abilities for Multi-modality Foundation Models paper page: huggingface.co/papers/2311.06… Multi-modality foundation models, as represented by GPT-4V, have brought a new paradigm for low-level visual perception and understanding tasks, that can…

Tweet Image 1

Chaofeng Reposted

Our evaluation datasets for Q-Bench huggingface.co/papers/2309.14… are released on huggingface! Specifically, the datasets are splitted into dev set (fully accessible) and test set (image+questions accessible, correct answer hidden), and the labels will also be released soon!


Chaofeng Reposted

Our team at Google Research is hiring student researcher for topics related to text-to-image synthesis and editing. Please write to me at kelvinckchan@google.com or drop me a DM if you are interested. (Retweets are welcomed!)


Chaofeng Reposted

We're pleased to see our #CodeFormer is integrated into #stablediffusion webui 🥳🥳 Try out our demos on github.com/sczhou/CodeFor… to fix your AI-generated arts. #stablediffusionart #aiart #aiartist

Tweet Image 1
Tweet Image 2

If you’re running a local instance of #stablediffusion WebUI, highly recommend you go check this github every few days as it’s constantly updated. For example, it now has both GFPGAN and CodeFormer builtin as face fixing options. github.com/AUTOMATIC1111/…



Chaofeng Reposted

Our newest system DALL·E 2 can create realistic images and art from a description in natural language. See it here: openai.com/dall-e-2/


Chaofeng Reposted

This is just wild. #StyleGAN3


Loading...

Something went wrong.


Something went wrong.