File size: 1,099 Bytes
4bd3bd1
 
 
 
a75b5ce
 
4bd3bd1
 
 
 
a75b5ce
4bd3bd1
 
 
 
582737d
 
4bd3bd1
 
 
a75b5ce
4bd3bd1
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
---
license: mit
size_categories:
- 1K<n<10K
tags:
- Audio-Visual-Event-Localization
---

## AVE Dataset

The original AVE dataset ported from the [GitHub](https://github.com/YapengTian/AVE-ECCV18/)

## Notes

- One video may contain different audio-visual events, so the total number of videos is not 4143.
- annotations.txt: annotations of the AVE dataset. For each sample, you can find its event category, YouTube ID, Quality (all good, meaning that it contains an AVE), start time of an audio-visual event, and the end time of an audio-visual event.
- train/val/test-Set.txt: training/validation/testing set used in the original paper.

## Citation

If you use the AVE dataset, please consider citing the original paper:

```
@inproceedings{TianECCV2018,
	 title={Audio-Visual Event Localization in Unconstrained Videos},
     author={Tian, Yapeng and Shi, Jing and Li, Bochen and Duan, Zhiyao and Xu, Chenliang},
	 editor="Ferrari, Vittorio and Hebert, Martial and Sminchisescu, Cristian and Weiss, Yair",
	 booktitle="Computer Vision -- ECCV 2018",
	 year="2018",
	 publisher="Springer",
}
```