└── README.md /README.md: -------------------------------------------------------------------------------- 1 | # AVA Actions Dataset 2 | The AVA dataset densely annotates 80 atomic visual actions in 430 movie clips with actions localized in space and time, resulting in 1.62M action labels with multiple labels per human occurring frequently. Clips are drawn from 15-minute contiguous segments of movies, to open the door for temporal reasoning about activities. The dataset is split into 235 videos for training, 64 videos for validation, and 131 videos for test. This page aims to provide the download instructions and mirror sites for AVA Dataset. Please visit the [project page](https://research.google.com/ava/) for more details on the dataset. 3 | ## Download Videos 4 | CVDF hosts the videos in the AVA dataset. Please download the videos with the url patterns: 5 | ``` 6 | https://s3.amazonaws.com/ava-dataset/trainval/[file_name] 7 | https://s3.amazonaws.com/ava-dataset/test/[file_name] 8 | ``` 9 | You can download the list of training/validation file names [here](https://s3.amazonaws.com/ava-dataset/annotations/ava_file_names_trainval_v2.1.txt), and the test filenames [here](https://s3.amazonaws.com/ava-dataset/annotations/ava_file_names_test_v2.1.txt). 10 | 11 | ## Download Annotations 12 | The public annotations, for the training and validation sets, can be downloaded here [ava_v2.2.zip](https://s3.amazonaws.com/ava-dataset/annotations/ava_v2.2.zip). 13 | 14 | # AVA ActiveSpeaker Dataset 15 | AVA ActiveSpeaker associates speaking activity with a visible face, on the AVA v1.0 videos, resulting in 3.65 million frames labeled across ~39K face tracks. A detailed description of this dataset is in the [arXiv paper](https://arxiv.org/abs/1901.01342). 16 | 17 | ## Download Videos 18 | CVDF hosts the videos in AVA ActiveSpeaker. Please download the videos with the url patterns: 19 | ``` 20 | https://s3.amazonaws.com/ava-dataset/trainval/[file_name] 21 | ``` 22 | You can download the list of file names [here](https://s3.amazonaws.com/ava-dataset/annotations/ava_speech_file_names_v1.txt). 23 | 24 | ## Download Annotations 25 | The public annotations for AVA ActiveSpeaker can be downloaded here [ava_activespeaker_train_v1.0.tar.bz2](https://s3.amazonaws.com/ava-dataset/annotations/ava_activespeaker_train_v1.0.tar.bz2) and [ava_activespeaker_val_v1.0.tar.bz2](https://s3.amazonaws.com/ava-dataset/annotations/ava_activespeaker_val_v1.0.tar.bz2). 26 | 27 | 28 | # AVA Speech Dataset 29 | The AVA-Speech dataset densely annotates speech activity for the movie clips in the AVA v1.0 dataset. It explicitly labels 3 background noise conditions (Clean Speech, Speech with background Music, and Speech with background Noise), resulting in ~40K labeled segments spanning 40 hours of data. Please visit the [project page](https://research.google.com/ava/) for more details on the dataset. 30 | 31 | ## Download Videos 32 | CVDF hosts the videos in AVA Speech. Please download the videos with the url patterns: 33 | ``` 34 | https://s3.amazonaws.com/ava-dataset/trainval/[file_name] 35 | ``` 36 | You can download the list of file names [here](https://s3.amazonaws.com/ava-dataset/annotations/ava_speech_file_names_v1.txt). 37 | 38 | ## Download Annotations 39 | The public annotations for AVA speech can be downloaded here [ava_speech_labels_v1.csv](https://s3.amazonaws.com/ava-dataset/annotations/ava_speech_labels_v1.csv). 40 | --------------------------------------------------------------------------------