nvidia-holoscan/holohub
Mar 2026 – Mar 2026
1 Month active
Languages Used
Python
Technical Skills
3D ReconstructionComputer VisionDockerMachine LearningPython
Vishwesh
PROFILE
Vishwesh
Developed an end-to-end surgical scene reconstruction pipeline for the nvidia-holoscan/holohub repository, leveraging Python, Docker, and advanced computer vision techniques. The solution integrated real-time Holoscan streaming with parallel depth and segmentation models, enabling metric depth and pose estimation for clinical scenarios. A five-phase pipeline was orchestrated through a single entry script supporting headless and resumable runs, facilitating CI and server automation. The work included deformable Gaussian Splatting training for non-rigid scenes, comprehensive documentation, and reproducible Docker-based deployment. This release provided a robust, automated workflow for surgical scene reconstruction, with detailed packaging and contributor credits to support ongoing development.
Overall Statistics
Feature vs Bugs
100%Features
Repository Contributions
1Total
Bugs
0
Commits
1
Features
1
Lines of code
10,799
Activity Months1
Your Network
1666 peopleSame Organization
@nvidia.com1624
Shared Repositories
42
Work History
March 2026
1 Commits • 1 Features
Mar 1, 2026March 2026 performance summary for nvidia-holoscan/holohub: Delivered an end-to-end surgical scene reconstruction pipeline using Gaussian Splatting with real-time Holoscan integration and Docker-based deployment. Implemented a five-phase pipeline (inference, depth/segmentation/pose estimation, dataset handling, training, and live viewer) orchestrated via a single entry script (run_gsharp.py) with headless and resume capabilities for CI/server automation. Real-time Phase 1 processing uses Holoscan to run Depth Anything V2 and MedSAM3 in parallel with a live 3-panel preview, enabling metric depth and pose estimation (VGGT) for clinical applicability. Also added deformable GSplat training for non-rigid scenes, and comprehensive documentation and packaging (README, DEV_NOTES, assets guide). Docker-first workflow supports reproducible deployments and CI integration. Release v0.2 with contributor credits.
1 Commits • 1 Features
Mar 1, 2026March 2026 performance summary for nvidia-holoscan/holohub: Delivered an end-to-end surgical scene reconstruction pipeline using Gaussian Splatting with real-time Holoscan integration and Docker-based deployment. Implemented a five-phase pipeline (inference, depth/segmentation/pose estimation, dataset handling, training, and live viewer) orchestrated via a single entry script (run_gsharp.py) with headless and resume capabilities for CI/server automation. Real-time Phase 1 processing uses Holoscan to run Depth Anything V2 and MedSAM3 in parallel with a live 3-panel preview, enabling metric depth and pose estimation (VGGT) for clinical applicability. Also added deformable GSplat training for non-rigid scenes, and comprehensive documentation and packaging (README, DEV_NOTES, assets guide). Docker-first workflow supports reproducible deployments and CI integration. Release v0.2 with contributor credits.
March 2026
Activity
Loading activity data...
Quality Metrics
Correctness100.0%
Maintainability80.0%
Architecture100.0%
Performance80.0%
AI Usage80.0%
Skills & Technologies
Programming Languages
Python
Technical Skills
3D ReconstructionComputer VisionDockerMachine LearningPython
Repositories Contributed To
Overview of all repositories you've contributed to across your timeline