PhD candidate @ Technical University of Munich
Munich, Germany
- Tweets 101
- Following 672
- Followers 859
- Likes 1,683
Dave Zhenyu Chen retweeted
📢 OneCanvas: 3D Scene Understanding via Panoramic Reprojection
We extract features from video frames and reproject them into one occlusion-free view of the whole scene that a 2D VLM reads just like a normal image. We can center this view on any viewpoint, including an agent's own pose for situated reasoning.
The same projection lets us create spatial training tasks with no human annotation, solvable only by reasoning over the 3D positions of real object features placed on an otherwise empty canvas.
The result is a stock 2D VLM that reasons in 3D, setting a new state of the art across spatial benchmarks at far less compute.
🌐 baranowskibrt.github.io/onec…
▶️ piped.video/NIaHLB9gA7s
Great work by @baranowskibrt & @davech2y
1
78
488
42,808
Dave Zhenyu Chen retweeted
Delighted to share the #CVPR 2024 paper compilation of our TUM Visual Computing Lab & collaborators:
piped.video/POH-CW1uyhg
Amazing work by everyone - so proud about it. It's going to be a fantastic conference in #Seattle this week.
Looking forward to seeing everyone!
1
21
113
13,184
Dave Zhenyu Chen retweeted
📢 Excited to Share Our New Survey on 3D-LLMs!
This work is a collaboration between researchers from @UniofOxford, @TU_Muenchen, @KAUST_News, @mbzuai, @SFU, and @ETH_en.
📄Paper: arxiv.org/pdf/2405.10255
🔗Project: github.com/ActiveVisionLab/A…
#AI #LLM #3D
🧵
5
24
62
17,916
I’m super happy to be invited to this wonderful research group. If you’re interested in cutting-edge research in 3D vision, come and join us!
Introducing our team members at #AnySyn3D! Explore our recent works with our collaborative partners. If you’re interested in connecting or partnering with us, please don’t hesitate to reach out via email at anysyn3d@gmail.com.
website: anysyn3d.github.io
#AIGC #AI #3D
1
31
4,669
👨🎓 What a wonderful journey at VC & AI Group at @TU_Muenchen! HUGE thank you to my PhD advisor @MattNiessner, my co-advisor Angel Chang, all my collaborators, the PhD committee members Rüdiger and @moElhoseiny, and all my colleagues! THANK YOU 🙏
Congrats to @davech2y for passing his PhD defense 🎉🥳🍾
Dave's PhD thesis "Grounding Natural Language to 3D Scenes" makes major contributions in 3D scene understanding with natural language, including 3D localization, object detection, and 3D representation learning from text.
10
2
113
11,289
Happy to announce that SceneTex has been accepted at #CVPR2024🎉
Combined with 3D geometry generative models like
@yawarnihal's MeshGPT, SceneTex can deliver fully textured 3D scenes!😉
Huge congrats to the team: Haoxuan Li @hyjameslee @SergeyTulyakov @MattNiessner
6
37
254
27,696
Project: daveredrum.github.io/SceneTe…
Code: github.com/daveredrum/SceneT…
Paper: arxiv.org/abs/2311.17261
1
7
1,301
Proudly present SceneTex!
If you can describe the style, you can texture the 𝐰𝐡𝐨𝐥𝐞 𝐬𝐜𝐞𝐧𝐞.
Check out our paper, code, and results -> daveredrum.github.io/SceneTe…
(1/2)
📢📢𝗦𝗰𝗲𝗻𝗲𝗧𝗲𝘅 📢📢
Given scene geometry and text prompt -> SceneTex generates high-quality textures.
Main idea: directly optimize scene texture with gradients from a score-distillation objective with view sampling.
piped.video/watch?v=-x8oMeLo…
daveredrum.github.io/SceneTe…
2
1
34
5,158
We’ll be presenting Text2Tex at #ICCV2023 - come check out our cool results!!! 😎
⏰ Oct 6, 10:30 am - 12:30 pm
📍 Room “Foyer Sud” - 091
(1/2)
Happy to announce that Text2Tex has been accepted at #ICCV2023 🎉
Taking a mesh and a text prompt as input, Text2Tex generates high quality textures - it's fully automated and easy to scale to many models!
Project: daveredrum.github.io/Text2Te…
Video: piped.video/2ve8tJ9LlcA
1
3
19
6,758
The 3rd Workshop on Language for 3D Scenes will take place tomorrow afternoon at #ICCV2023 @ICCVConference
Come and join us to check out the cool research on 3D vision-language 😁
📆 Oct 3, 2023
⏰ 14:00 - 17:50 Paris time
📍 W03
More info: languagefor3dscenes.github.i…
1
4
24
3,974
Check out the cool projects from our lab - looking forward to #ICCV2023 !!!
Excited to share the #ICCV 2023 paper compilation of our TUM Visual Computing Lab & collaborators: piped.video/u6IVsyKPyQ0
Thanks so much everyone for all the hard work. It's going to be an awesome conference - really looking forward to Paris next week!
5
1,462
Huge congrats Prof. @JustusThies 🎉🎉
🎉 Willkommen Prof. Justus Thies an der #TUDarmstadt! Mit seiner Expertise in 3D Graphics & Vision eröffnen sich spannende Perspektiven für Forschung und Lehre. Einblick in sein faszinierendes Forschungsgebiet? Hier entlang: tu-darmstadt.de/universitaet… @CS_TUDarmstadt @JustusThies
8
1,093
Congrats Justus! I’m so happy for you!
I will join @TUDarmstadt as full professor for 3D Graphics & Vision in Sept.
I am very thankful for the support from all my collaborators and students! I am privileged to have outstanding people around me, as my mentors and collaborators, and I keep on learning from them!
1
3
943
We open-sourced Text2Tex! If you can describe it, you can texture it! Have fun texturing objects with unlimited possibilities ^^
Paper: arxiv.org/abs/2303.11396
Project: daveredrum.github.io/Text2Te…
Code: github.com/daveredrum/Text2T…
(1/2)
Happy to announce that Text2Tex has been accepted at #ICCV2023 🎉
Taking a mesh and a text prompt as input, Text2Tex generates high quality textures - it's fully automated and easy to scale to many models!
Project: daveredrum.github.io/Text2Te…
Video: piped.video/2ve8tJ9LlcA
2
34
103
29,209
2/2 of my leading-author papers were accepted at #ICCV2023 🎉🎉🎉 Huge congrats to all our collaborators! See you in Paris!!!
Super happy that our group has eight papers accepted at #ICCV'2023!
Fantastic topics: 3D generative models, diffusion, NeRFs, neural rendering, and many more - check it out!
Very excited about all works! Congrats to all students and collaborators :)
niessnerlab.org/publications…
3
37
6,571
Multi-view images strike back! 3D-CLR enables reasoning and learning in 3D scenes from 2D vision-language priors - #L3DS (Language for 3D scenes) is really striving!!! 🔥🔥🔥
Check out this fantastic project from Yining! Connecting languages to 3D scenes has been on hype 🔥🔥. From ScanRefer (by @davech2y) to SQA3D(sqa3d.github.io), now with this new dataset you can even do it with multi-view pictures! This is indeed pivotal to embodied AI!
6
1,115
Dave Zhenyu Chen retweeted
Great tips! One small tip to add here: Talk to current and graduated students from that lab about their experiences.
How to choose a good Ph.D. advisor?
Grad school application season is coming to an end. Congrats on the offers! Now, how would you choose your advisor that will help you succeed?
Some tips below (that I wish I knew).
1
12
2,624