Sadegh Aliakbarian

Scientist @Microsoft

[Sadegh Aliakbarian @aa_sadegh]

I’ll be presenting DAViD (https://t.co/rZI417afvX) at #ICCV2025 next week! If you’re around, make sure to stop by our poster at Exhibit Hall I on Tue 21 Oct from 3:15-5:15p.m to chat about it!

#SyntheticData #HumanCentricAI #EfficientAI #EthicalAI #OpenScience

[Sadegh Aliakbarian @aa_sadegh]

@jhugestar Very interesting work! Congrats!
And, what are the odds, we’re going to present (another) DAViD at ICCV: https://t.co/rZI417afvX 😊

[Fatemeh Saleh @fatemeh_saleh]

Excited to share our latest work at #SIGGRAPHAsia2024!

Look Ma, no markers: Holistic performance capture without the hassle

https://t.co/CcD6WJINeN

[Sadegh Aliakbarian @aa_sadegh]

@artsiom_s @finkd It’s indeed pretty wild 🤯 great work @artsiom_s

[Amir Habibian @amir_habibian]

A summary of our recent research on efficient diffusion models for images, videos and 3D:
https://t.co/HRqoPVFasB

arXivs:
- https://t.co/TcKm9Khv0i
- https://t.co/0EFdC6dP1U
- https://t.co/kjHVaHtx6i
- https://t.co/nXKy6NHsE5

#GenerativeAI #diffusion #efficient

[Amir Habibian @amir_habibian]

Five fellowships (each 40,000$) available for PhD students in European universities.

More info:
https://t.co/v2UGqWR9mc

DM if have any question.

[Sadegh Aliakbarian @aa_sadegh]

Checkout "3DiFACE", our new generative model for Speech-driven 3D facial animation and editing 👇

[Sadegh Aliakbarian @aa_sadegh]

@cs_jiaxi_jiang @fatemeh_saleh @dopomoc Definitely, local 6D + joint pos is essential. Loss on global rotation is very very important though.

[Sadegh Aliakbarian @aa_sadegh]

"HMD-NeMo: Online 3D Avatar Motion Generation from Sparse Observations" will be presented at #ICCV2023. Please come and meet us (Room "Foyer Sud" - 013, on Thursday 10:30AM)

Link to paper: https://t.co/oxy2gEXeem

Together with @fatemeh_saleh, David, Pashmina, and @dopomoc

[Sadegh Aliakbarian @aa_sadegh]

@cs_jiaxi_jiang @fatemeh_saleh @dopomoc That's a good question, while I don't have enough experiment to back it up, but computing the additional loss on global joint transformation matrices typically boosts the performance a lot. Euc distance for position and geodesic loos for rot mat (in 6D) is very effective

[Sadegh Aliakbarian @aa_sadegh]

@cs_jiaxi_jiang @fatemeh_saleh @dopomoc 3/3 Happy to discuss more if you're attending ICCV :)

[Sadegh Aliakbarian @aa_sadegh]

@cs_jiaxi_jiang @fatemeh_saleh @dopomoc 2/n in addition to that (2) our SE(3) loss is contributing a lot when it comes to MPJPE. See fig (9) that removing this loss term leads to a model with MPJPE of >5cm. Finally (3) optimization on top of NeMo's output (table 6/7) yields further improvement.

[Mohsen Ghafoorian @mohsen_gh87]

* 3D Distillation: Improving Self-Supervised Monocular Depth Estimation on Reflective Surfaces
@jujihong
@ukrdailo
The following gives a short summary of these two papers:
https://t.co/ASeL0WIvhs

[Sadegh Aliakbarian @aa_sadegh]

@mohsen_gh87 @jujihong @ukrdailo Looking forward to meeting you 😊

[Mohammad Mahdi (Medi) Derakhshani @mmderakhshani]

Thrilled to announce that I will be presenting my previous internship paper at #ICCV2023 in Paris next Thursday! Most work done during an internship at Samsung AI Cambridge. Work w/ @ESanchezLozano, @AdrianBulat, @victorturrisi, @cgmsnoek, @tzimiro, Brais Martinez.

1/2

[Sadegh Aliakbarian @aa_sadegh]

Check out Imitator, a novel approach to generating personalised speech-driven facial animations.

Come to our #ICCV2023 poster on Friday (Room "Nord" - 058) 😊

[Justus Thies @JustusThies]

Next week @ICCVConference, @balathambiraja will present his work Imitator which learns personalized speech-driven 3D facial animation.

Web : https://t.co/xtDdlTi9eP
Code: https://t.co/0SscNZplOH

In collaboration with Ikhsanul Habibie, @aa_sadegh, @dopomoc, Christian Theobalt

[Sadegh Aliakbarian @aa_sadegh]

@sitzikbs Thanks for having me on the podcast!

[Peyman Milanfar @docmilanfar]

This ad is amazing - it challenges your assumptions and is technically impressive. Watch to the end.

https://t.co/613V07Yq1k