Tianshu Kuai (@TianshuKuai) profile picture

CS PhD Student @UMontreal and @Mila_Quebec | MSc and BASc @UofT

Montreal
Joined July 2022
Photo by Tianshu Kuai (@TianshuKuai)Photo by Tianshu Kuai (@TianshuKuai)Photo by Tianshu Kuai (@TianshuKuai)Photo by Tianshu Kuai (@TianshuKuai)Photo by Tianshu Kuai (@TianshuKuai)Photo by Tianshu Kuai (@TianshuKuai)Photo by Tianshu Kuai (@TianshuKuai)Photo by Tianshu Kuai (@TianshuKuai)Photo by Tianshu Kuai (@TianshuKuai)Photo by Tianshu Kuai (@TianshuKuai)
I'll be recruiting PhD and MSc students through Mila - consider applying if you want to work at the intersection of machine learning and 3D geometry!
Mila's annual supervision request process opens on October 15 to receive MSc and PhD applications for Fall 2025 admission! Join our community! More information here mila.quebec/en/prospective-s…
2
36
2
98
18,151
Mila's annual supervision request process opens on October 15 to receive MSc and PhD applications for Fall 2025 admission! Join our community! More information here mila.quebec/en/prospective-s…
3
77
51
164
368,103
Tianshu Kuai retweeted
📢🔍 Super excited to present Spatially Aware Multiview Diffusers (SPAD) at #CVPR2024! SPAD enables 3D consistent multi-view image generation from text or image inputs. It is trained using a high-quality Objaverse subset on 32 H100s! Code & Paper links at the end! 🧵👇
2
15
2
69
6,890
Tianshu Kuai retweeted
1/ Excited to share our #CVPR2024 work LEOD! We propose a label-efficient learning framework for object detection with event cameras, which performs on par with SOTA models with **10x fewer labels**! Paper: arxiv.org/abs/2311.17286 Code: github.com/Wuziyi616/LEOD
2
12
1
33
3,930
Tianshu Kuai retweeted
🚀 Excited to share our SIGGRAPH Asia work! 📚 iNVS: Repurposing Diffusion Inpainters for Novel View Synthesis 🔍TLDR: Maximize reuse of source image pixels for novel view synthesis from a single image. 🔗 Website: yashkant.github.io/invs 🌐 #SIGGRAPHAsia2023 #Research #iNVS
7
15
2
96
13,925
Tianshu Kuai retweeted
🧵If you are attending ICCV🇫🇷🥐, make sure to check Tristan (@TristanAA97) presenting our work, "Reference-guided Controllable Inpainting of Neural Radiance Fields". website: ashmrz.github.io/reference-g… abs: arxiv.org/abs/2304.09677 Paper summary in the thread🧵.
1
6
2
46
16,003
Excited to share that our work on Object-centric learning with Diffusion Models has been accepted as a spotlight at NeurIPS! 💡🎉 Huge kudos to my student @Dazitu_616 for leading that work. It's been a great collaboration involving @JingyuHu16, @WuyueLu_Ivy, and @animesh_garg.
1/ Excited to share our NeurIPS’23 Spotlight paper SlotDiffusion! Our work on diffusion-based object-centric learning enables unsupervised segmentation, visual generation, and downstream video prediction & VQA, all with one model! Project, paper, code: slotdiffusion.github.io
1
6
29
5,947
Tianshu Kuai retweeted
1/ Excited to share our NeurIPS’23 Spotlight paper SlotDiffusion! Our work on diffusion-based object-centric learning enables unsupervised segmentation, visual generation, and downstream video prediction & VQA, all with one model! Project, paper, code: slotdiffusion.github.io
3
18
4
119
32,944
I am excited to share that our paper "Self-Supervised Image-to-Point Distillation via Semantically Tolerant Contrastive Loss" has been accepted to #CVPR2023! @TRoboAIL Paper: arxiv.org/abs/2301.05709
7
20
1,189
Super excited to share that my first project as a PhD student was accepted at CVPR 2023! #CVPR2023 😄 We show how to recover accurate camera poses from as few as 3 images! More details to come! w/ @jasonyzhang2 @taiyasaki @igilitschenski @DaveLindell arxiv.org/abs/2211.16991
27
113
14
717
100,171
Our paper DVF "Dense Voxel Fusion for 3D Object Detection" was accepted at #WACV2023 ! DVF densely fuses image predictions and LiDAR voxel features improving expressiveness in low point density regions. @TRoboAIL arxiv.org/abs/2203.00871
3
12
Tianshu Kuai retweeted
[1/7] 🥁🍽️Attending #ECCV2022🚀, check out LaTeRF🌟, a framework for extracting photo-realistic 3D object assets from real-world scenes with minimal annotation🧵! Paper📜: arxiv.org/abs/2207.01583 Code🧑‍💻: github.com/ashmrz/LaTeRF Video▶️: youtu.be/rRgsLZoyHek
1
9
1
43
Tianshu Kuai retweeted
With enough 3D data, inherently 3D models will outperform many 2D tasks. Until then, distilling 2D information into 3D is almost the best we can do🌟. Also checkout our #ECCV2022 work LaTeRF, which extends this idea for object extraction🚀. Code: github.com/ashmrz/LaTeRF
Semantic-NeRF! Super simple, just add semantic outputs to a NeRF network and you can label a full 3D scene from highly sparse or noisy annotations. #ICCV2021
3
13
Tianshu Kuai retweeted
🏠🧹Introducing Housekeep, a benchmark to evaluate commonsense reasoning in embodied AI agents. Appearing at #ECCV2022! Website: yashkant.github.io/housekeep… Paper: arxiv.org/abs/2205.10712 Data in colab: tinyurl.com/3sdetm7r 🧵
3
24
3
81
Congrats @ashmrz10! I am very excited about this work. In some sense, this is also the first paper coming out of my lab.
Thanks @_akhaliq for sharing our #ECCV2022 work. We propose LaTeRF to extract 3D object representations from 2D input images with a minimal set of supervision; User clicks are used to remove the background/foreground, while a text prompt is utilized to resolve the occlusions.
1
2
21