Junsong Chen

I am currently a research intern at NVIDIA Research supervised by Dr. Enze Xie and Prof. Song Han. I was also a Research Assistant at The University of Hong Kong and a PhD Candidate at DLUT. During my PhD time, I have been fortunate to work under the supervision of Prof. Ping Luo, and Prof. Huchuan Lu.

Email  /  CV  /  Google Scholar  /  Github

Research Interests

In 2023&2024, I participate in the following research topics:

• AIGC, e.g. Diffusion Models in 2D/3D
• Large Language Models (LLMs), e.g.
• Autonomous Driving, e.g. 3D perception & prediction

Previously, I did research on 2D/3D detection, segmentation and object tracking. Representative works include MetaBEV for 3D objection detection and segmentation and ARKitTrack for 2D object tracking.

News

• [06/2024] One paper was accepted by ECCV2024.
• [01/2024] One paper was accepted by ICLR2024 (Spotlight).
• [12/2023] One paper was accepted by AAAI2023.
• [07/2023] One paper was accepted by ICCV2023.
• [03/2023] One paper was accepted by CVPR2023.

Selected Publications

For the latest full list please check here.

AIGC (Text-to-Image Diffusion)

PixArt-α is a Transformer-based T2I diffusion model whose image generation quality is competitive with state-of-the-art image generators (e.g., Imagen, SDXL, and even Midjourney), and the training speed markedly surpasses existing large-scale T2I models, e.g., PixArt-α only takes 12% of Stable Diffusion v1.5's training time (753 vs. 6,250 A100 GPU days).

In this paper, we introduce PixArt-Σ, a Diffusion Transformer model (DiT) capable of directly generating images at 4K resolution. PixArt-Σ represents a significant advancement over its predecessor, PixArt-α, offering images of markedly higher fidelity and improved alignment with text prompts.

Perception (object detection/segmentation/tracking)

MetaBEV is a 3D Bird-Eye's View (BEV) perception model that is robust to sensor missing/failure, supporting both single modality mode (camera/lidar) and multi-modal fusion mode with strong performance.

ARKitTrack, a new RGB-D track-ing dataset for both static and dynamic scenes captured by consumer-grade LiDAR scanners equipped on Apple’s iPhone and iPad. ARKitTrack contains 300 RGB-D sequences, 455 targets, and 229.7K video frames in total. Along with the bounding box annotations and frame-level attributes.