We are excited to be present our #ICCV2025 paper today! Stop by our poster - very happy to share our findings and hear your thoughts! 🗓 Tuesday 21st October 2025 🕥 03:00 PM to 05:00 PM (Hawaii Time) 📍 Hawaii Convention Center 🏁 353 (1/N)

For people filling Sec. 1.1 (Compulsory) of the form, type the following commands: # CPU info egrep "model name" /proc/cpuinfo | head -1 egrep -c ^processor /proc/cpuinfo # GPU info nvidia-smi # Memory info htop # Check Memory lsblk -d -o name,rota # 0 is SSD / 1 is HDD

My role at Meta's SAM team (MSL, previously at FAIR Perception) has been impacted within 3 months of joining after PhD. If you work with multimodal LLMs for grounding or complex reasoning, or have a long-term vision of unified understanding and generation, let's talk. I am on

CHARM3R improves generalization to unseen camera heights by more than 45%, achieving SoTA performance on the CARLA dataset. (11/N)

To mitigate the effect of camera height changes, we propose Camera Height Robust Monocular 3D Detector (CHARM3R), which averages both depth estimates within the model. (10/N)

While ground-based methods perform worse in detection, we mathematically prove and also empirically observe contrastive OoD trends in mean depth error of regressed and ground-based depth models, respectively, under camera height changes. (9/N)

while there exists an alternative ground-based depth (8/N)

Most SoTA models are regressed models, (7/N)

Why is this problem challenging? (6/N)

Monocular 3D object detectors, while effective on data from one ego camera height, struggle with unseen or out-of-distribution camera heights. (5/N)

Joint work with amazing team of @33yuliangguo (Bosch Research North America), Zhihao Zhang (MSU), Xinyu Huang (Bosch Research North America), Liu Ren (Bosch Research North America) and Xiaoming Liu (MSU) (4/N)

Code+Models+Training scripts: https://t.co/l9aJaGTi85 Paper: https://t.co/0jFC6atZBg Talk: https://t.co/oMbhkieMET TLDR: 💪CHARM3R closes 45% of the gap, while the baselines perform nearly zero. (3/N)

We tackle the challenge of making monocular 3D detectors work on unseen camera height data (bots, trucks) by training only on single car height data. In other words, we study Out of Distribution regression problem. (2/N)

📢The @neur_reps Workshop @NeurIPSConf 2024 extends submission deadline. Consider submitting your research on incorporating physical laws, symmetry, geometry, or topology in neural nets to this exciting venue.

1page Resume

2Page Resume: https://t.co/MZiBiUN1e6

I'm officially on the industry job market! I am looking for full time Research Scientist,  Research Engineer and Applied Scientist positions starting January 2025. I like working on 3D problems like 3D detection and 3D reconstruction. If you find my profile interesting, DM!

We are thrilled to be presenting our #CVPR2024 paper today! Stop by our poster - we're eager to share our findings and hear your thoughts! 🗓 Thursday 20th June 2024 🕥 10:30 AM to 12:00 noon (Pacific Time) 📍 Exhibit Hall (Room 4A-E), Seattle Arch Convention Center 🏁 58

particularly for large objects (> 8m in length). 13/n

and improves existing detectors on the nuScenes leaderboard 12/n