Awesome Open Source
Awesome Open Source

Lightly Logo

GitHub Unit Tests codecov

Lightly is a computer vision framework for self-supervised learning.

We, at Lightly, are passionate engineers who want to make deep learning more efficient. That's why - together with our community - we want to popularize the use of self-supervised methods to understand and curate raw image data. Our solution can be applied before any data annotation step and the learned representations can be used to visualize and analyze datasets. This allows to select the best core set of samples for model training through advanced filtering.


Lightly offers features like

  • modular framework
  • support for multi-gpu training using PyTorch Lightning
  • easy to use and written in a PyTorch like style
  • supports custom backbone models for self-supervised pre-training

Supported Models


Want to jump to the tutorials and see lightly in action?

Tutorials of using the lightly packge together with the Lightly Platform:

Community and partner projects:

Quick Start

Lightly requires Python 3.6+. We recommend installing Lightly in a Linux or OSX environment.


  • hydra-core>=1.0.0
  • numpy>=1.18.1
  • pytorch_lightning>=1.0.4
  • requests>=2.23.0
  • torchvision
  • tqdm


You can install Lightly and its dependencies from PyPI with:

pip3 install lightly

We strongly recommend that you install Lightly in a dedicated virtualenv, to avoid conflicting with your system packages.

Lightly in Action

With lightly you can use latest self-supervised learning methods in a modular way using the full power of PyTorch. Experiment with different backbones, models and loss functions. The framework has been designed to be easy to use from the ground up.

import torch
import torchvision
import lightly.models as models
import lightly.loss as loss
import as data

# the collate function applies random transforms to the input images
collate_fn = data.ImageCollateFunction(input_size=32, cj_prob=0.5)

# create a dataset from your image folder
dataset = data.LightlyDataset(input_dir='./my/cute/cats/dataset/')

# build a PyTorch dataloader
dataloader =
    dataset,                # pass the dataset to the dataloader
    batch_size=128,         # a large batch size helps with the learning
    shuffle=True,           # shuffling is important!
    collate_fn=collate_fn)  # apply transformations to the input images

# use a resnet backbone
resnet = torchvision.models.resnet.resnet18()
resnet = nn.Sequential(*list(resnet.children())[:-1])

# build the simclr model
model = models.SimCLR(resnet, num_ftrs=512)

# use a criterion for self-supervised learning
criterion = loss.NTXentLoss(temperature=0.5)

# get a PyTorch optimizer
optimizer = torch.optim.SGD(model.parameters(), lr=1e-0, weight_decay=1e-5)

You can easily use another model like SimSiam by swapping the model and the loss function.

# build the simsiam model
model = models.SimSiam(resnet, num_ftrs=512)

# use the SimSiam loss function
criterion = loss.SymNegCosineSimilarityLoss()

Use PyTorch Lightning to train the model:

trainer = pl.Trainer(max_epochs=max_epochs, gpus=1)

Or train the model on 4 GPUs:

trainer = pl.Trainer(

Command-Line Interface

Lightly is accessible also through a command-line interface (CLI). To train a SimCLR model on a folder of images you can simply run the following command:

lightly-train input_dir=/mydataset

To create an embedding of a dataset you can use:

lightly-embed input_dir=/mydataset checkpoint=/mycheckpoint

The embeddings with the corresponding filename are stored in a human-readable .csv file.


Currently implemented models and their accuracy on cifar10 and imagenette. All models have been evaluated using kNN. We report the max test accuracy over the epochs as well as the maximum GPU memory consumption. All models in this benchmark use the same augmentations as well as the same ResNet-18 backbone. Training precision is set to FP32 and SGD is used as an optimizer with cosineLR. One epoch on cifar10 takes ~35 seconds on a V100 GPU. Learn more about the cifar10 and imagenette benchmark here


Model Epochs Batch Size Test Accuracy
MoCo 800 256 0.827
SimCLR 800 256 0.847
SimSiam 800 256 0.827
BarlowTwins 800 256 0.801
BYOL 800 256 0.851


Model Epochs Batch Size Test Accuracy
MoCo 200 128 0.83
SimCLR 200 128 0.78
SimSiam 200 128 0.73
BarlowTwins 200 128 0.84
BYOL 200 128 0.85
MoCo 200 512 0.85
SimCLR 200 512 0.83
SimSiam 200 512 0.81
BarlowTwins 200 512 0.78
BYOL 200 512 0.84
MoCo 800 128 0.89
SimCLR 800 128 0.87
SimSiam 800 128 0.80
MoCo 800 512 0.90
SimCLR 800 512 0.89
SimSiam 800 512 0.91


Below you can see a schematic overview of the different concepts present in the lightly Python package. The terms in bold are explained in more detail in our documentation.

Overview of the lightly pip package

Next Steps

Head to the documentation and see the things you can achieve with Lightly!


To install dev dependencies (for example to contribute to the framework) you can use the following command:

pip3 install -e ".[dev]"

For more information about how to contribute have a look here.

Running Tests

Unit tests are within the tests folder and we recommend to run them using pytest. There are two test configurations available. By default only a subset will be run. This is faster and should take less than a minute. You can run it using

python -m pytest -s -v

To run all tests (including the slow ones) you can use the following command.

python -m pytest -s -v --runslow

Code Linting

We provide a Pylint config following the Google Python Style Guide.

You can run the linter from your terminal either on a folder

pylint lightly/

or on a specific file

pylint lightly/

Further Reading

Self-supervised Learning:


  • Why should I care about self-supervised learning? Aren't pre-trained models from ImageNet much better for transfer learning?

    • Self-supervised learning has become increasingly popular among scientists over the last year because the learned representations perform extraordinarily well on downstream tasks. This means that they capture the important information in an image better than other types of pre-trained models. By training a self-supervised model on your dataset, you can make sure that the representations have all the necessary information about your images.
  • How can I contribute?

    • Create an issue if you encounter bugs or have ideas for features we should implement. You can also add your own code by forking this repository and creating a PR. More details about how to contribute with code is in our contribution guide.
  • Is this framework for free?

    • Yes, this framework completely free to use and we provide the code. We believe that we need to make training deep learning models more data efficient to achieve widespread adoption. One step to achieve this goal is by leveraging self-supervised learning. The company behind lightly commited to keep this framework open-source.
  • If this framework is free, how is the company behind lightly making money?

    • Training self-supervised models is only part of the solution. The company behind lightly focuses on processing and analyzing embeddings created by self-supervised models. By building, what we call a self-supervised active learning loop we help companies understand and work with their data more efficiently. This framework acts as an interface for our platform to easily upload and download datasets, embeddings and models. Whereas the platform will cost for additional features this frameworks will always remain free of charge (even for commercial use).


If you want to cite the framework feel free to use this:

  author={Igor Susmelj, Matthias Heller, Philipp Wirth, Jeremy Prescott, Malte Ebner et al.},
  journal={GitHub. Note:},

Get A Weekly Email With Trending Projects For These Topics
No Spam. Unsubscribe easily at any time.
Python (1,142,951
Machine Learning (31,745
Deep Learning (23,727
Pytorch (11,616
Computer Vision (8,220
Embeddings (629
Self Supervised Learning (344
Active Learning (195
Contrastive Learning (150
Related Projects