Skip to content

vincentschen/limited-label-scene-graphs

Repository files navigation

Scene Graph Prediction with Limited Labels

Overview

  • Scene graphs capture visual relationships between objects in images, e.g. <person, riding, bike>. They have emerged as useful in a number of computer vision tasks, including visual question answering and captioning.

  • However, most scene graph datasets are sparse due to annotator error!

  • This work attempts to overcome limitations of human annotators using a semi-supervised method, taking advantage of both limited labels and unlabeled data, to generate training datasets for scene graphs.

  • Approach:

    • We leverage image-agnostic features, which are cheap to extract given bounding box pairs of images.
    • With as few as n=10 labeled relationships per predicate (e.g. "walk", "ride", "eat"), we learn heuristics, shallow decision trees that can serve as noisy labelers.
    • Because these heuristics are error prone, we learn a generative model to combine and denoise the outputs of these heuristics, producing probabilistic labels for each object pair.
    • Using probabilistic labels over the unlabeled training data, we can bootstrap training for any downstream scene graph model!

Setup

Please run the following script to download the VisualGenome dataset.

./scripts/get_visualgenome.sh

To create the virtual environment with appropriate requirements:

virtualenv -p python3 .env
source .env/bin/activate
pip install -r requirements.txt

Demo

All instructions for this demonstration are included in main.ipynb.

Paper

You can find the full paper corresponding to this work, presented at ICCV 2019, at https://arxiv.org/abs/1904.11622.

Bibtex

Please refer to the following citation if you are building on this work:

@inproceedings{Chen_2019_ICCV,
    author = {Chen, Vincent S. and Varma, Paroma and Krishna, Ranjay and Bernstein, Michael and Re, Christopher and Fei-Fei, Li},
    title = {Scene Graph Prediction With Limited Labels},
    booktitle = {The IEEE International Conference on Computer Vision (ICCV)},
    month = {October},
    year = {2019}
}

Contributions

Feel free to open an issue or send an email if you have any questions!

About

Scene Graph Prediction with Limited Labels

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published