Initial code release

.github/workflows/lint_and_test.yaml

@@ -0,0 +1,31 @@
+name: Lint and Test
+on:
+  push:
+    branches:
+      - main
+  pull_request:
+    branches:
+      - main
+
+permissions:
+  contents: read
+jobs:
+  lint_and_test:
+    name: lint and test
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v3
+      - name: Install poetry
+        run: pipx install poetry
+      - name: Setup python and poetry
+        uses: actions/setup-python@v4
+        with:
+          python-version: 3.8.16
+          cache: poetry
+          cache-dependency-path: poetry.lock
+      - name: Install dependencies
+        run: poetry install
+      - name: Lint code
+        run: poetry run pre-commit run --all-files --show-diff-on-failure
+      - name: Run tests
+        run: poetry run pytest -v -m "not slow" -m "not dataset"

.gitignore

@@ -0,0 +1,166 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+.pybuilder/
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+#   For a library or package, you might want to ignore these files since the code is
+#   intended to run in multiple environments; otherwise, check them in:
+# .python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# poetry
+#   Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#   https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control
+#poetry.lock
+poetry.toml
+
+# pdm
+#   Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control.
+#pdm.lock
+#   pdm stores project-wide configurations in .pdm.toml, but it is recommended to not include it
+#   in version control.
+#   https://pdm.fming.dev/#use-with-ide
+.pdm.toml
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# PyCharm
+#  JetBrains specific template is maintained in a separate JetBrains.gitignore that can
+#  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
+#  and can be added to the global gitignore or merged into this file.  For a more nuclear
+#  option (not recommended) you can uncomment the following to ignore the entire idea folder.
+#.idea/
+
+.vscode/
+.ruff_cache
+logs/
+lightning_logs/

.pre-commit-config.yaml

@@ -0,0 +1,25 @@
+repos:
+-   repo: local
+    hooks:
+    -   id: isort
+        name: isort
+        entry: poetry run isort
+        language: system
+        types: [python]
+    -   id: black
+        name: black
+        entry: poetry run black
+        language: system
+        types: [python]
+    -   id: ruff
+        name: ruff
+        entry: poetry run ruff check --fix --show-source --show-fixes --exit-non-zero-on-fix .
+        language: system
+        types: [python]
+-   repo: https://github.com/pre-commit/pre-commit-hooks
+    rev: v4.3.0
+    hooks:
+    -   id: trailing-whitespace
+    -   id: end-of-file-fixer
+    -   id: check-yaml
+    -   id: check-added-large-files

LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2023 Maximilian Seitzer and Andrii Zadaianchuk
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

README.md

@@ -1,9 +1,80 @@
-# [Object-Centric Learning for Real-World Videos by Predicting Temporal Feature Similarities (NeurIPS 2023)](https://martius-lab.github.io/videosaur/)
+# VideoSAUR
+
+This is the code release for the paper **Object-Centric Learning for Real-World Videos by Predicting Temporal Feature Similarities (NeurIPS 2023)**, by Andrii Zadaianchuk, Maximilian Seitzer and Georg Martius.
+
+- Webpage: https://martius-lab.github.io/videosaur
+- Arxiv: https://arxiv.org/abs/2306.04829
+- OpenReview: https://openreview.net/forum?id=t1jLRFvBqm
+
 ![Temporal Feature Similarities](https://zadaianchuk.github.io/videosaur/static/images/sim_loss-1.png)
 
-## Citation 
+## Summary
+
+Unsupervised video-based object-centric learning is a promising avenue to learn structured representations from large, unlabeled video collections, but previous approaches have only managed to scale to real-world datasets in restricted domains. Recently, it was shown that the reconstruction of pre-trained self-supervised features leads to object-centric representations on unconstrained real-world image datasets. Building on this approach, we propose a novel way to use such pre-trained features in the form of a temporal feature similarity loss. This loss encodes semantic and temporal correlations between image patches and is a natural way to introduce a motion bias for object discovery. We demonstrate that this loss leads to state-of-the-art performance on the challenging synthetic MOVi datasets. When used in combination with the feature reconstruction loss, our model is the first object-centric video model that scales to unconstrained video datasets such as YouTube-VIS.
+
+## Usage
+
+### Setup
+
+First, setup the python environment setup. We use [Poetry](https://python-poetry.org/) for this:
+
+```
+poetry install
+```
+
+Then you could run a test configuration to see if everything works:
+
+```
+poetry run python -m videosaur.train tests/configs/test_dummy_image.yml
+```
+
+Second, to download the datasets used in this work, follow the instructions in [data/README.md](data/README.md).
+By default, datasets are expected to be contained in the folder `./data`.
+You can change this to the actual folder your data is in by setting the environment variable `VIDEOSAUR_DATA_PATH`, or by running `train.py` with the `--data-dir` option.
+
+### Training
+
+Run one of the configurations in `configs/videosaur`, for example:
+
+```
+poetry run python -m videosaur.train configs/videosaur/movi_c.yml
+```
+
+The results are stored in a folder created under the log root folder (by defaults `./logs`, changeable by the argument `--log-dir`).
+If you want to continue training from a previous run, you can use the `--continue` argument, like in the following command:
+
+```
+poetry run python -m videosaur.train --continue <path_to_log_dir_or_checkpoint_file> configs/videosaur/movi_c.yml
+```
+
+## Results
+
+### VideoSAUR
+
+We list the results you should roughly be able to obtain with the configs included in this repository:
+
+| Dataset      | Model Variant    |  Video ARI | Video mBO | Config                      |
+|--------------|------------------|------------|-----------|-----------------------------|
+| MOVi-C       | ViT-B/8, DINO    |  64.8      | 38.9      | videosaur/movi_c.yml        |
+| MOVi-E       | ViT-B/8, DINO    |  73.9      | 35.6      | videosaur/movi_e.yml        |
+| YT-VIS 2021  | ViT-B/16, DINO   |  39.5      | 29.1      | videosaur/ytvis.yml         |
+| YT-VIS 2021  | ViT-B/14, DINOv2 |  39.7      | 35.6      | videosaur/ytvis_dinov2.yml  |
+
+### DINOSAUR
+
+We also include a configuration for the DINOSAUR model from our previous paper [Bridging the gap to real-world object-centric learning](https://arxiv.org/abs/2209.14860).
+This configuration yields improved results compared to the DINOSAUR model in the original paper (mainly
+due to using DINOv2 pre-trained features).
+Note that there might be minor differences in the metrics, as the numbers here are computed for 224x224 masks, compared to 320x320 masks in the DINOSAUR paper.
+
+| Dataset     | Model Variant      | Image ARI | Image mBO | Config                             |
+|-------------|--------------------|-----------|-----------|------------------------------------|
+| COCO        | ViT-B/14, DINOv2   | 45.6      | 29.6      | dinosaur/coco_base14_dinov2.yml    |
+
+## Citation
+
+If you make use of this repository, please use the following bibtex entry to cite us:
 
-Please use the following bibtex entry to cite us:
 ```
   @inproceedings{zadaianchuk2023objectcentric,
       title={Object-Centric Learning for Real-World Videos by Predicting Temporal Feature Similarities},
@@ -12,3 +83,10 @@ Please use the following bibtex entry to cite us:
       year={2023},
   }
 ```
+
+## License
+
+This codebase is released under the MIT license.
+Some parts of the codebase were adapted from other codebases.
+A comment was added to the code where this is the case.
+Those parts are governed by their respective licenses.