Sam Xu

Xiang Xu (Sam)

I am a senior research scientist at Autodesk AI Lab, working on CAD generative models. Previously, I completed PhD/MSc in CS from Simon Fraser Univeristy, supervised by Prof. Yasutaka Furukawa. And BSc in ECE from Carnegie Mellon University, advised by Prof. Kris Kitani.

Education
  • PhD in CS, 2021 - 2024

    Simon Fraser University

  • MSc in CS, 2019 - 2021

    Simon Fraser University

  • BSc in ECE, 2014 - 2018

    Carnegie Mellon University

Experiences

Publications

See Google Scholar for full publications
BrepGen: A B-rep Generative Diffusion Model with Structured Latent Geometry

A diffusion-based generative approach that directly outputs a CAD B-rep. We represent a B-rep as a novel structured latent geometry tree format. B-rep topology is implicitly represented by node duplication.

Hierarchical Neural Coding for Controllable CAD Model Generation

Represent high-level CAD design concepts as a hierarchical tree of neural codes. User controls the generation or auto-completion of CAD models by specifying the target design using a code tree.

SkexGen: Autoregressive Generation of CAD Construction Sequences with Disentangled Codebooks

Using disentangled codebooks to generate diverse and high-quality CAD models, enhances user control, and enables efficient exploration of the CAD design space.

Structured Outdoor Architecture Reconstruction by Exploration and Classification

An explore-and-classify framework for building architectural reconstruction. Our method explores the structure space by heuristic modifications and classifing the correctness of updated results.

D3D-HOI: Dynamic 3D Human-Object Interactions from Videos

Monocular video dataset with ground truth annotations of 3D object pose, shape and part motion. We leverage 3D human pose for more accurate inference of the object spatial layout and dynamics.

MCMI: Multi-Cycle Image Translation with Mutual Information Constraints

Treat single-cycle image translation as modules that can be used recurrently where the process is bounded by mutual information constraints between the input and output images.

Error Correction Maximization for Deep Image Hashing

We use the Hamming bound to derive the optimal criteria for learning hash codes with a deep network.