/icevision

An Agnostic Computer Vision Framework - Pluggable to any Training Library: Fastai, Pytorch-Lightning with more to come

Primary LanguagePythonApache License 2.0Apache-2.0

logo

An Agnostic Computer Vision Framework


tests docs codecov PyPI version Downloads

black license Discord


IceVision is the first agnostic computer vision framework to offer a curated collection with hundreds of high-quality pre-trained models from Torchvision, Open MMLab's MMDetection, Ultralytic's YOLOv5, Ross Wightman's EfficientDet and soon PyTorch Image Models. It orchestrates the end-to-end deep learning workflow allowing to train networks with easy-to-use robust high-performance libraries such as PyTorch-Lightning and Fastai.

IceVision Unique Features:

  • Data curation/cleaning with auto-fix

  • Access to an exploratory data analysis dashboard

  • Pluggable transforms for better model generalization

  • Access to hundreds of neural net models

  • Access to multiple training loop libraries

  • Multi-task training to efficiently combine object detection, segmentation, and classification models

Installation

pip install icevision[all]

For more installation options, check our docs.

Important: We currently only support Linux/MacOS.

Quick Example: How to train the Fridge Objects Dataset

Open In Colab image

image

Happy Learning!

If you need any assistance, feel free to:

Join our Forum