typo
Browse files
README.md
CHANGED
|
@@ -1,6 +1,6 @@
|
|
| 1 |
# Dataset: MultiVSR
|
| 2 |
|
| 3 |
-
We introduce a large-scale multilingual lip-reading dataset: MultiVSR. The dataset comprises a total of
|
| 4 |
|
| 5 |
<p align="center">
|
| 6 |
<img src="dataset_teaser.gif" alt="MultiVSR Dataset Teaser">
|
|
|
|
| 1 |
# Dataset: MultiVSR
|
| 2 |
|
| 3 |
+
We introduce a large-scale multilingual lip-reading dataset: MultiVSR. The dataset comprises a total of 12,000 hours of video footage, covering English + 12 non-English languages. MultiVSR is a massive dataset with a huge diversity in terms of the speakers as well as languages, with approximately 1.6M video clips across 123K YouTube videos. Please check the [https://www.robots.ox.ac.uk/~vgg/research/multivsr/](website) for samples.
|
| 4 |
|
| 5 |
<p align="center">
|
| 6 |
<img src="dataset_teaser.gif" alt="MultiVSR Dataset Teaser">
|