diff --git a/README.md b/README.md index 64b5806..918539b 100644 --- a/README.md +++ b/README.md @@ -8,6 +8,7 @@ https://asanakoy.github.io/cliquecnn/ * The paper can be downloaded from https://arxiv.org/abs/1608.08792 * Labels that we gathered for Olympic Sports can be found in [olympic_sports_retrieval/data](olympic_sports_retrieval/data) +* Bounding boxes for the OlympicSports dataset [olympic_sports_retrieval/data/bboxes.tar.gz](olympic_sports_retrieval/data/bboxes.tar.gz) * All our pretrained models for Olympic Sports dataset can be downloaded from [here](https://hcicloud.iwr.uni-heidelberg.de/index.php/s/kRp6b454Dd0wnts) * Caffe's deploy file: [olympic_sports_retrieval/models/deploy.prototxt](olympic_sports_retrieval/models/deploy.prototxt) * Evaluation script for Olympic Sports: [olympic_sports_retrieval/calculate_roc_auc.py](olympic_sports_retrieval/calculate_roc_auc.py) diff --git a/olympic_sports_retrieval/data/bboxes.tar.gz b/olympic_sports_retrieval/data/bboxes.tar.gz new file mode 100644 index 0000000..61d1893 Binary files /dev/null and b/olympic_sports_retrieval/data/bboxes.tar.gz differ diff --git a/olympic_sports_retrieval/data/bboxes_README.txt b/olympic_sports_retrieval/data/bboxes_README.txt new file mode 100644 index 0000000..3b629fa --- /dev/null +++ b/olympic_sports_retrieval/data/bboxes_README.txt @@ -0,0 +1,5 @@ +Every *.bb file corresponds to a video sequence. +Every line in *.bb file defines a bounding box for the specific frame. + +*.bb file has the following structure: +frame_index xmin ymin xmax ymax