What’s new in computer vision research from Lambda?

At Lambda, we help researchers move from breakthrough ideas to large-scale deployment with the right tools and infrastructure.

Two highlights from our contributions to ICCV conference:

DepR: Depth Guided Single-view Scene Reconstruction with Instance-level Diffusion: A new way to turn 2D images into realistic 3D scenes using depth and instance-level diffusion, paving the path toward scalable 3D reconstruction.

Video-MMLU: A Massive Multi-Discipline Lecture Understanding Benchmark: A pioneering benchmark for multimodal reasoning that enables models to learn from and reason over complex video data.

Awarded Outstanding Paper at ICCV’s Multimodal Reasoning Workshop.

View our demos and papers:

DepR → Demo & Paper

Video-MMLU → Demo & Paper