Ruiyuan GAO
@ruiyuan_gao
Followers
27
Following
56
Media
6
Statuses
27
PhD student at CUHK | Interested in AI security, generative models for autonomous driving. Open to take jobs starting in Fall 2025
Sha Tin District, Hong Kong
Joined July 2016
🎉 Got accepted by ICCV2025. We renamed to MagicDrive-V2 with new features integrated. Stay tuned to our GitHub repository. Source code coming soon!
Introducing MagicDriveDIT, the upgraded MagicDrive with DiT architecture—still in control! Now supporting high-res, 848x1600x6 perspective autonomous driving videos, 20 seconds long at 12 FPS. Check it out! 🚗✨ Paper: https://t.co/VMczRkJPw4 Website: https://t.co/K6qjmkdalP
0
1
1
🕹️We are excited to introduce "ChronoEdit: Towards Temporal Reasoning for Image Editing and World Simulation" ChronoEdit reframes image editing as a video generation task to encourage temporal consistency. It leverages a temporal reasoning stage that denoises with “video
6
38
141
We are excited to share Cosmos-Drive-Dreams 🚀 A bold new synthetic data generation (SDG) pipeline powered by world foundation models—designed to synthesize rich, challenging driving scenarios at scale. Models, Code, Dataset, Tookit are released. Website:
11
48
107
Just released 2 checkpoints for high-res images (up to 424x800) and 1 for long videos (6x60 frames)! All MagicDrive CKPTs are available! Enjoy the open source!
Very happy to share that our MagicDrive has been accepted by #ICLR2024! MagicDrive: Street View Generation with Diverse 3D Geometry Control Paper: https://t.co/hVhuia1swe Project page: https://t.co/c4bMGbDybO The code is coming soon! Stay tuned.
0
0
0
Model checkpoint released now. Enjoy!
Code is available at: https://t.co/9592aDwZ09 Models coming soon!🤗
0
0
0
Code is available at: https://t.co/9592aDwZ09 Models coming soon!🤗
github.com
[ICCV 2025] Official implementation of the paper “MagicDrive-V2: High-Resolution Long Video Generation for Autonomous Driving with Adaptive Control” - flymin/MagicDrive-V2
Introducing MagicDriveDIT, the upgraded MagicDrive with DiT architecture—still in control! Now supporting high-res, 848x1600x6 perspective autonomous driving videos, 20 seconds long at 12 FPS. Check it out! 🚗✨ Paper: https://t.co/VMczRkJPw4 Website: https://t.co/K6qjmkdalP
0
0
0
Introducing MagicDriveDIT, the upgraded MagicDrive with DiT architecture—still in control! Now supporting high-res, 848x1600x6 perspective autonomous driving videos, 20 seconds long at 12 FPS. Check it out! 🚗✨ Paper: https://t.co/VMczRkJPw4 Website: https://t.co/K6qjmkdalP
0
1
1
Both tracks of our challenge launch now! https://t.co/CKw2Nq327X Please register before making your submission!
coda-dataset.github.io
Multimodal Perception and Comprehension of Corner Cases in Autonomous Driving, ECCV 2024 Workshop
🚗Ready to revolutionize #AutonomousDriving with #Multimodal learning? Join our @eccvconf Workshop on Multimodal Perception & Comprehension of Corner Cases in Autonomous Driving! ✨Don't miss out – let's construct reliable driving agents! Check https://t.co/mXNntRg7n6
#ECCV2024
0
0
0
We have released the 16-frame weight and 60-frame config for MagicDrive video generation! The new code for MagicDrive3D will be located at https://t.co/lowur1rArn Stay tuned!
github.com
Official implementation of the paper “MagicDrive3D: Controllable 3D Generation for Any-View Rendering in Street Scenes” - flymin/MagicDrive3D
Don't miss out MagicDrive3D if you're interested in MagicDrive! Our latest endeavor introduces controllable autonomous driving 3D scene generation. project page: https://t.co/eew1lsjtZN One step closer to a generative simulator!
0
1
0
Checkout our W-CODA workshop at ECCV2024 now!
🚗Ready to revolutionize #AutonomousDriving with #Multimodal learning? Join our @eccvconf Workshop on Multimodal Perception & Comprehension of Corner Cases in Autonomous Driving! ✨Don't miss out – let's construct reliable driving agents! Check https://t.co/mXNntRg7n6
#ECCV2024
0
0
0
Don't miss out MagicDrive3D if you're interested in MagicDrive! Our latest endeavor introduces controllable autonomous driving 3D scene generation. project page: https://t.co/eew1lsjtZN One step closer to a generative simulator!
3
1
2
In Vienna 🇦🇹 and I will present our work MagicDrive: Street View Generation with Diverse 3D Geometry Control. If you are interested in diffusion models for autonomous driving, please come to the presentation or DM me!! @iclr_conf #ICLR2024
0
0
5
Happy to share that our paper, DetDiffusion, has been accepted by #CVPR2024! With so many papers proposing generative models and perception models that can help each other, our DetDiffusion, for the first time, boosts the synergy of the two. ArXiv: https://t.co/cw65uOrbHZ
0
0
0
Happy to share that our paper, MMA-Diffusion, has been accepted by #CVPR2024! Prompt Filters or Safety Checkers for T2I services are vulnerable to adversarial attacks. We expect new defensive methods to protect T2I. ArXiv: https://t.co/T0xWoOzUlN Code: https://t.co/bXCfr0KuJS
0
0
0
The code for video generation is now online. Check it out: https://t.co/vg5Xge843C Among all street view generations/world models for autonomous driving, MagicDrive is still the 1st and ONLY open-sourced work!
github.com
[ICLR24] Official implementation of the paper “MagicDrive: Street View Generation with Diverse 3D Geometry Control” - cure-lab/MagicDrive
Very happy to share that our MagicDrive has been accepted by #ICLR2024! MagicDrive: Street View Generation with Diverse 3D Geometry Control Paper: https://t.co/hVhuia1swe Project page: https://t.co/c4bMGbDybO The code is coming soon! Stay tuned.
0
0
1
Training code is now ready! Check out at https://t.co/vg5Xge843C Among all street view generations / world models for autonomous driving, MagicDrive is the first and only open-sourced work!
github.com
[ICLR24] Official implementation of the paper “MagicDrive: Street View Generation with Diverse 3D Geometry Control” - cure-lab/MagicDrive
Very happy to share that our MagicDrive has been accepted by #ICLR2024! MagicDrive: Street View Generation with Diverse 3D Geometry Control Paper: https://t.co/hVhuia1swe Project page: https://t.co/c4bMGbDybO The code is coming soon! Stay tuned.
0
0
0
Demo code with pretrained model is released at
github.com
[ICLR24] Official implementation of the paper “MagicDrive: Street View Generation with Diverse 3D Geometry Control” - cure-lab/MagicDrive
Very happy to share that our MagicDrive has been accepted by #ICLR2024! MagicDrive: Street View Generation with Diverse 3D Geometry Control Paper: https://t.co/hVhuia1swe Project page: https://t.co/c4bMGbDybO The code is coming soon! Stay tuned.
0
0
0
Very happy to share that our MagicDrive has been accepted by #ICLR2024! MagicDrive: Street View Generation with Diverse 3D Geometry Control Paper: https://t.co/hVhuia1swe Project page: https://t.co/c4bMGbDybO The code is coming soon! Stay tuned.
0
0
4