PolarNick239
Follow
Stars
[NeurIPS 2024] Direct3D: Scalable Image-to-3D Generation via 3D Latent Diffusion Transformer
Demonstrates seven different techniques for order-independent transparency in Vulkan.
[CVPR 2025] FoundationStereo: Zero-Shot Stereo Matching
Free and open source library for AI object detection and semantic segmentation in geospatial rasters. 🚀
Code release for "3D reconstruction with fast dipole sums"
Computer Vision course for CS bachelors in UCU (2019)
[CVPR 2024] GLACE: Global Local Accelerated Coordinate Encoding
Sample benchmark demonstrating the VK_KHR_cooperative_matrix extension
[MMM‘24 Oral]CT-MVSNet: Efficient Multi-View Stereo with Cross-scale Transformer
[CVPR'24]🦿GoMVS: Geometrically Consistent Cost Aggregation for Multi-View Stereo
A simple open-source disk benchmark tool for Linux distros
CIFAR-10 speedruns: 94% in 2.6 seconds and 96% in 27 seconds
Train CIFAR10 to 94% accuracy in a few minutes/seconds. Based on https://github.com/davidcpage/cifar10-fast
Train to 94% on CIFAR-10 in <6.3 seconds on a single A100. Or ~95.79% in ~110 seconds (or less!)
Efficient Edge-Preserving Multi-view Stereo Network for Depth Estimation, AAAI2023
SurfelWarp: Efficient Non-Volumetric Dynamic Reconstruction
Implementation of Newcombe et al. CVPR 2015 DynamicFusion paper
Randomized Correspondence Algorithm for Structural Image Editing
UE5's Nanite implementation using WebGPU. Includes the meshlet LOD hierarchy, software rasterizer and billboard impostors. Culling on both per-instance and per-meshlet basis.
[ECCV 2024] - Improving Zero-shot Generalization of Learned Prompts via Unsupervised Knowledge Distillation
Trim 3D Gaussian Splatting for Accurate Geometry Representation
A 2D Gaussian Splatting paper for no obvious reasons. Enjoy!
[SIGGRAPH'24] 2D Gaussian Splatting for Geometrically Accurate Radiance Fields
Few-shot NeRF by Adaptive Rendering Loss Regularization (ECCV 2024)
Tessendorf FFT based ocean waves and buoyancy in Godot 4 using compute shaders
[NeurIPS 2024] Depth Anything V2. A More Capable Foundation Model for Monocular Depth Estimation
Single source file BC1-5 and BC7 encoders and BC1-5/7 decoders with MIT or Public Domain licenses