CV

Education

April 2022 - Present: Saarland University
- M.Sc. in Visual Computing.
- Projects:
  - Computer Graphics: Lightwave - Ray Tracer, first prize in the rendering competition.
  - High-Level Computer Vision: A damper for the overfitting problem in training INR for Image Denoising.
Aug 2017 – Aug 2021: Ho Chi Minh city University of Technology
- B.E. in Computer Science (GPA: 8.58 / 10.0).
- Bachelor Thesis: Image Denoising for Smartphone Cameras.
  - Implemented and empirically optimized an encoder-decoder CNN architecture
  - Quality: PSNR ≈ 39 in SIDD and DND (SOTA performance)
  - Speed: 12MP/s on Snap 855 (× 100 times faster than other SOTA methods)

Oct 2022 - Present: Research Assistant in MPI-INF
- Novel View Synthesis.
Oct 2021 - April 2022: Research Engineer at Graphicsminer
- SLAM (Simultaneous Localization and Mapping) for smartphones.
- Real-Time 3D Tracking
  - Applied Gauss-Newton algorithm to optimize a region-based level-set energy function w.r.t a 6DoF object pose parameterized by a twist coordinate (Demo: Single-Object Tracking, Multiple-Object Tracking).
- Real-time Multiple-view Monocular Camera 3D Reconstruction.
  - Applied continuous max-flow algorithm to optimize a Bayesian inference-based energy function w.r.t a voxel grid shape representation (Demo).
  - Accelerated the algorithm with GLSL.
Oct. 2020 – Oct. 2021: Undergraduate Research Assistant at HCMUT
- Self-supervised CNN-based Image Denoising
  - Analysed Stein’s unbiased risk estimate and blind spot networks’ architectures
- Neural Image Denoising for Smartphones
  - Proposed and implemented a denoising model with 12MB/s processing speed on edge devices
Jun 2020 - Oct 2020: Intern at Graphicsminer
- Real-time video stabilization (reached 11 Fps on Pixel 1).