Vision Transformer Python Packages

mlx-vlm

MLX-VLM is a package for inference and fine-tuning of Vision Language Models (VLMs) on your Mac using MLX.

mmdet

OpenMMLab Detection Toolbox and Benchmark

mmcls

OpenMMLab Pre-training Toolbox and Benchmark

mmpretrain

OpenMMLab Pre-training Toolbox and Benchmark

pix2tex

pix2tex: Using a ViT to convert images of equations into LaTeX code.

ko-hand-ocr

Korean handwriting OCR — a 36M-param model trained from scratch on synthetic data only. No handwriting dataset, no inherited license terms. CPU-only.

towhee

Towhee is a framework that is dedicated to making neural data processing pipelines simple and fast.

canvit-pytorch

CanViT: Toward Active-Vision Foundation Models (NeurIPS 2026). Model, pretraining, task specialization and evaluation code.

mist-medical

MIST: A simple and scalable end-to-end framework for 3D medical imaging segmentation.

mambavision

[CVPR 2025] Official PyTorch Implementation of MambaVision: A Hybrid Mamba-Transformer Vision Backbone

attention-and-transformers

Transformers goes brrr... Attention and Transformers from scratch in TensorFlow. Currently contains Vision transformers, MobileViT-v1, MobileViT-v2, MobileViT-v3

fastervit

[ICLR 2024] Official PyTorch implementation of FasterViT: Fast Vision Transformers with Hierarchical Attention

garmentiq

Free & Open Source. Precise and flexible garment measurements from images - no tape measures, no delays, just fashion - forward automation.

misfit-medical

MISFIT: A simple, scalable, end-to-end framework for pretraining 3D medical imaging foundation models using masked autoencoders (MAE)

canvit-mlx

CanViT: Toward Active-Vision Foundation Models (NeurIPS 2026). Model, pretraining, task specialization and evaluation code.

supreme-unlearning

Multi-GPU framework for reproducible image-unlearning evaluation, with independent training, unlearning and evaluation seeds and extensible datasets, models, methods and metrics. Published at WIPE-OUT 2 (ECML-PKDD 2026).

pyrafuse

DINOv3-based semantic segmentation for skin, fabric, and background — PyTorch, Hugging Face, and TensorRT.

pai-easycv

An all-in-one toolkit for computer vision

canvit

CanViT: Toward Active-Vision Foundation Models (NeurIPS 2026). Model, pretraining, task specialization and evaluation code.

canvit-nnx

CanViT: Toward Active-Vision Foundation Models (NeurIPS 2026). Model, pretraining, task specialization and evaluation code.

canvit-core

CanViT: Toward Active-Vision Foundation Models (NeurIPS 2026). Model, pretraining, task specialization and evaluation code.

fast-lightonocr

High-performance Rust inference engine for LightOnOCR-2-1B using ONNX Runtime.

tfimm

TensorFlow port of PyTorch Image Models (timm) - image models with pretrained weights.

towhee-models

Towhee is a framework that is dedicated to making neural data processing pipelines simple and fast.