Skip to content

Latest commit

 

History

History
112 lines (84 loc) · 3.4 KB

AVA.md

File metadata and controls

112 lines (84 loc) · 3.4 KB
Raw

简体中文 | English

AVA Data Preparation

This document mainly introduces the preparation process of AVA dataset. It mainly includes five parts: Video Data Download, Prepare Annotations, Cut video files, Extract the RGB frames, Pulling Proposal Files,et al. Before we start, please make sure that the directory is located at $PaddleVideo/data/ava/script.

1. Video data Download

For basic dataset information, you can refer to the official website AVA. For the dataset download, you can refer to the AVA Download ， which introduce the way to download the dataset. We also provide the shell script for downloading the video files

bash download_videos.sh

Furthermore,considering the difficulty in downloading, we upload the video files to Baidu cloud disk in the form of zip packages, and users can download it by themselves according to their needs. Link coming soon.

Note: the video files should be placed in data/ava/videos

2.Prepare Annotations

Next, you can run the following script to prepare annotations.

bash download_annotations.sh

This command will download ava_v2.1.zip for AVA v2.1 annotation. If you need the AVA v2.2 annotation, you can try the following script.

VERSION=2.2 bash download_annotations.sh

Note: In fact,we will also provide the annotation zip files in Baidu cloud disk

3. cut video files

Cut each video from its 15th to 30th minute and make them at 30 fps.

bash cut_videos.sh

4. Extract RGB Frames

you can use the ffmpeg to extract RGB frames by the following script.

bash extract_rgb_frames.sh

5.Pulling Proposal Files

The scripts are adapted from FAIR's Long-Term Feature Banks.

Run the follow scripts to fetch pre-computed proposal list.

bash fetch_ava_proposals.sh

6.Folder Structure

After the whole data pipeline for AVA preparation. you can get the rawframes (RGB), videos and annotation files for AVA.

In the context of the whole project (for AVA only), the folder structure will look like:

PaddleVideo
├── configs
├── paddlevideo
├── docs
├── tools
├── data
│   ├── ava
│   │   ├── annotations
│   │   |   ├── ava_dense_proposals_train.FAIR.recall_93.9.pkl
│   │   |   ├── ava_dense_proposals_val.FAIR.recall_93.9.pkl
│   │   |   ├── ava_dense_proposals_test.FAIR.recall_93.9.pkl
│   │   |   ├── ava_train_v2.1.csv
│   │   |   ├── ava_val_v2.1.csv
│   │   |   ├── ava_train_excluded_timestamps_v2.1.csv
│   │   |   ├── ava_val_excluded_timestamps_v2.1.csv
│   │   |   ├── ava_action_list_v2.1_for_activitynet_2018.pbtxt
│   │   ├── videos
│   │   │   ├── 053oq2xB3oU.mkv
│   │   │   ├── 0f39OWEqJ24.mp4
│   │   │   ├── ...
│   │   ├── videos_15min
│   │   │   ├── 053oq2xB3oU.mkv
│   │   │   ├── 0f39OWEqJ24.mp4
│   │   │   ├── ...
│   │   ├── rawframes
│   │   │   ├── 053oq2xB3oU
|   │   │   │   ├── img_00001.jpg
|   │   │   │   ├── img_00002.jpg
|   │   │   │   ├── ...