Annotate better with CVAT, the industry-leading data engine for machine learning. Used and trusted by teams at any scale, for data of any scale.

Official repo for "Mini-Gemini: Mining the Potential of Multi-modality Vision Language Models"

A new approach to API documentation in Sphinx.

EfficientViT is a new family of vision models for efficient high-resolution vision.

cryptography is a package designed to expose cryptographic primitives and recipes to Python developers.

🤓 Important machine learning knowledge, each article deeply analyzes theoretical knowledge

The Triton Inference Server provides an optimized cloud and edge inferencing solution.

Multithreaded Python without the GIL

Let us control diffusion models!

Using Low-rank adaptation to quickly fine-tune diffusion models.

Efficient and Diverse Batch Acquisition for Deep Bayesian Active Learning.

CV-CUDA™ is an open-source, GPU accelerated library for cloud-scale image processing and computer vision.

High Resolution Style Transfer in PyTorch with Color Control and Mixed Precision 🎨

Software that can perform photorealistic style transfer without the need of any post-processing steps.

Code to reproduce the results from the paper "Controlling Perceptual Factors in Neural Style Transfer" (https://arxiv.org/abs/1611.07865).

code for "Fast Patch-based Style Transfer of Arbitrary Style".

The source code of 'Visual Attribute Transfer through Deep Image Analogy'.

Generate image analogies using neural matching and blending.

A numpy implementation of the paper "Image Quilting for Texture Synthesis and Transfer" (2001)

Official Pytorch code for "Texture Reformer: Towards Fast and Universal Interactive Texture Transfer" (AAAI 2022)

《动手学深度学习》：面向中文读者、能运行、可讨论。中英文版被70多个国家的500多所大学用于教学。

Official Implementation of SynthTIGER (Synthetic Text Image Generator), ICDAR 2021

AITemplate is a Python framework which renders neural network into high performance CUDA/HIP C++ code. Specialized for FP16 TensorCore (NVIDIA GPU) and MatrixCore (AMD GPU) inference.

Model Quantization Benchmark

AugMix: A Simple Data Processing Method to Improve Robustness and Uncertainty

Corruption and Perturbation Robustness (ICLR 2019)

Discover augmentation strategies tailored for your dataset

PRIME: A Few Primitives Can Boost Robustness to Common Corruptions

TRADES (TRadeoff-inspired Adversarial DEfense via Surrogate-loss minimization)