ashawkey/kiuikit

A niche toolkit for computer vision and more.

What it solves

Kiuikit is a specialized toolkit designed to simplify common tasks in computer vision, with a particular focus on 3D vision. It provides a collection of utilities for handling 3D meshes, video/image processing, and neural network components that are often tedious to implement from scratch.

How it works

The toolkit operates as both a command-line interface (CLI) and a Python library. The CLI provides tools for 3D mesh viewing, super-resolution, and system/job management. The Python library offers modules for torch-native 3D mesh handling, vector and quaternion math, camera utilities, and pre-built PyTorch neural network blocks such as DiT and UNet.

Who it’s for

It is intended for researchers and developers working in computer vision and 3D reconstruction, who need efficient tools for data I/O, geometric math, and building AI models.

Highlights

  • 3D Vision Tools: Includes a GUI 3D mesh viewer (kire), torch-native mesh loading/exporting, and quaternion math.
  • Neural Network Blocks: Provides standalone PyTorch blocks for MLP, DiT, UNet (2D/3D), VAE, and flow matching.
  • Media I/O: High-performance video and image reading/writing with support for float/HDR and numpy/torch.
  • Developer Utilities: Includes a CUDA-synchronized timer, hash-grid encoding with CUDA backend, and LPIPS perceptual loss.

Related

  • Project
    facebookresearch/pytorch3d

    A PyTorch-based library for 3D Computer Vision research providing differentiable rendering and efficient tools for manipulating 3D meshes and point clouds.

  • Project
    TorchIO-project/torchio

    A Python library for PyTorch that provides tools for the efficient loading, preprocessing, and augmentation of 3D medical images.

  • Project
    pytorch/vision

    A PyTorch library providing popular datasets, model architectures, and image transformations for computer vision.

  • Project
    nagadomi/nunif

    A collection of PyTorch-based tools for image super-resolution, 2D-to-3D video conversion for VR, and image quality assessment for dataset filtering.

  • Project
    adalca/neurite

    A neural networks toolbox for medical image analysis that provides specialized tensor operations, prebuilt architectures, and plotting tools for spatial data.