Jia-Bin Huang Profile picture
Apr 2 4 tweets 3 min read Twitter logo Read on Twitter
Excited to share our #CVPR2023 on synthesizing new views along a camera trajectory from a **single image**!

How?
💡 The good old epipolar constraints in a pose-guided diffusion model!

Paper: arxiv.org/abs/2303.17598
Project: poseguided-diffusion.github.io
How does it work?

We train a diffusion model conditioned on 1) relative camera pose and 2) source image via a cross attention layer.

BUT, we don’t need to attend every location!

Fantastic correspondences and where to find them?

Epipolar geometry! ImageImage
How well does it work?

Surprisingly, this simple method achieves both higher per-frame quality AND better temporal consistency! Image
Super excited about this work and the future exploration on blending geometric inductive bias and powerful data-driven approaches!

Work brought to you by the AMAZING @HungYuTseng, Qinbo Li, @_ChangilKim, Suhib Alsisan, @JPKopf

Come chat with us at #CVPR2023!

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Jia-Bin Huang

Jia-Bin Huang Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @jbhuang0604

Apr 4
How to write math in a paper?

Math allows you to convey your idea precisely and concisely. But how to write them clearly? 🤔

Check out some high-level tips (with examples). 🧵
*Make it readable*

Math writing blends both NATURAL and MATH languages. It should be *readable*. Image
*Follow good style*

• texts in math: \mathrm for
• transpose: use ^\top instead of ^T
• big parentheses: use \left and \right Image
Read 11 tweets
Mar 21
How to write the Introduction?

As a junior student, writing the introduction of a research paper is arguably the most daunting part of paper writing. 😱

Here is a simple template I find useful:

3 Figures 🖼️ + 5 Questions 🤔
*🖼️ WHAT figure*

Start with the paper with a WHAT figure (known as the "teaser").

This figure shows only two things.
1⃣ Input
2⃣Output

This helps the readers understand what your work is about. Here are some of my favorite examples.
WHAT Example 1:

Mask R-CNN showcases its key results upfront without waiting until the result section.

1⃣ Input: Single images
2⃣Output: Instance segmentation masks Image
Read 20 tweets
Feb 22
How to decide what to work on?

So many exciting ideas, so little time! How can one narrow down to a specific idea to work on? 🤔

Sharing some strategies I found useful. 🧵
*Find your position*

Understand your unfair advantages over others, what the community cares about, and how you can best contribute.

Avoid crowding the ball (aka chasing the hottest problem in the field).
*Identify intrinsic motivation*

Work on problems and real applications that you genuinely care about.
Read 7 tweets
Feb 19
How to find research opportunities?

Finding opportunities to gain experience is arguably the most challenging part for students wishing to pursue grad school, particularly for those who don't have resources/connections.

Some tips on approaching potential mentors. 🧵
*Lead with value*

❌ "I am very interested ... I want to grow ... I want to get the experiences... I have many prior experiences ... me, me, me, me, me, me, me!"

✅ "Here is what I can contribute to your team. I can do this bc my experience/skills on blah blah blah."
*Go the extra mile*

❌ Generic email sent to 500 professors. "I found your project X inspiring."

✅ "I ran the code from your latest project X. Here are some limitations I found. I think we can explore this and that direction to tackle the problem."
Read 7 tweets
Feb 17
How to create a good table?

While in grad school, I thought my job writing the paper was done after dumping all the numerical numbers from my experiments in a table. 🤦‍♂️

Check out some tips that will help you improve the quality of your tables! 🧵
*1⃣ Avoid vertical lines*

Having vertical lines in a table almost always makes the table less readable. Avoid them at all costs.
*2⃣ Never, ever use double rules*

Feel the urge to use double rules (\hline)?

Try the \toprule, \midrule, \bottomrule using the booktabs package instead! It levels up your table in no time!
Read 8 tweets
Dec 6, 2022
How to write a reference letter for graduate admission?

After writing and reviewing many recommendation letters, I gradually learned what makes a good letter.

Here are some tips for people new to writing recommendation letters for the upcoming application season! 🧵
*It's your job*

If the student works closely with you on research projects, it's your RESPONSIBILITY to write the letter.

Don't ask students to write a complete draft so you can just sign your name.
*History of interaction*

Describe how you got to know the applicant, for how long, and how well can you comment on their potential?
Read 9 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us on Twitter!

:(