Kornia Geometric Computer Vision Library
What it is
Kornia is a differentiable geometric computer vision library built on PyTorch. It provides over 1,200 operators for image processing, camera calibration, and feature matching with support for batched parameters and GPU acceleration.
The problem it solves
Standard libraries like torchvision lack per-sample batched parameters and comprehensive geometric algorithms (e.g., camera models, minimal solvers), often forcing slow CPU-based OpenCV or scikit-image workflows.
What you could use it for
- Building differentiable image transformation pipelines
- Performing batched geometric estimation like homography and RANSAC
- Applying GPU-accelerated local feature detection and matching
- Creating annotated data augmentation workflows for 2D and 3D data
Why it matters
Significantly accelerates geometric computer vision tasks on GPUs and enables end-to-end differentiable training of vision models.
Read the source ↗Back to the library →
Summary written by an AI model from the source; check the source before you rely on it.