Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,74 @@
|
|
1 |
-
|
2 |
-
|
3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
````markdown
|
2 |
+
---
|
3 |
+
license: cc-by-nc-4.0
|
4 |
+
---
|
5 |
+
|
6 |
+
# ResponseNet
|
7 |
+
|
8 |
+
**ResponseNet** is a large-scale dyadic video dataset designed for **Online Multimodal Conversational Response Generation (OMCRG)**. It fills the gap left by existing datasets by providing high-resolution, split-screen recordings of both speaker and listener, separate audio channels, and word‑level textual annotations for both participants.
|
9 |
+
|
10 |
+
## Paper
|
11 |
+
|
12 |
+
If you use this dataset, please cite:
|
13 |
+
|
14 |
+
> **ResponseNet: A High‑Resolution Dyadic Video Dataset for Online Multimodal Conversational Response Generation**
|
15 |
+
> *Authors: Luo, Cheng and Wang, Jianghui and Li, Bing and Song, Siyang and Ghanem, Bernard*
|
16 |
+
|
17 |
+
|
18 |
+
## Features
|
19 |
+
|
20 |
+
- **696** temporally synchronized dyadic video pairs (over **14 hours** total).
|
21 |
+
- **High-resolution** (1024×1024) frontal‑face streams for both speaker and listener.
|
22 |
+
- **Separate audio channels** for fine‑grained verbal and nonverbal analysis.
|
23 |
+
- **Word‑level textual annotations** for both participants.
|
24 |
+
- **Longer clips** (average **73.39 s**) than REACT2024 (30 s) and Vico (9 s), capturing richer conversational exchanges.
|
25 |
+
- **Diverse topics**: professional discussions, emotionally driven interactions, educational settings, interdisciplinary expert talks.
|
26 |
+
- **Balanced splits**: training, validation, and test sets with equal distributions of topics, speaker identities, and recording conditions.
|
27 |
+
|
28 |
+
## Data Fields
|
29 |
+
|
30 |
+
Each example in the dataset is a dictionary with the following fields:
|
31 |
+
|
32 |
+
- `video/speaker`: Path to the speaker’s video stream (1024×1024, frontal view).
|
33 |
+
- `video/listener`: Path to the listener’s video stream (1024×1024, frontal view).
|
34 |
+
- `audio_speaker`: Path to the speaker’s separated audio channel.
|
35 |
+
- `audio/listener`: Path to the listener’s separated audio channel.
|
36 |
+
- `transcript/speaker`: Word‑level transcription for the speaker (timestamps included).
|
37 |
+
- `transcript/listener`: Word‑level transcription for the listener (timestamps included).
|
38 |
+
- `vector/speaker`: Path to the speaker’s facial attributes.
|
39 |
+
- `vector/listener`: Path to the listener’s facial attributes.
|
40 |
+
|
41 |
+
## Dataset Splits
|
42 |
+
|
43 |
+
We follow a standard **6:2:2** split ratio, ensuring balanced distributions of topics, identities, and recording conditions:
|
44 |
+
|
45 |
+
| Split | # Video Pairs | Proportion (%) |
|
46 |
+
|------------|---------------|----------------|
|
47 |
+
| **Train** | 417 | 59.9 |
|
48 |
+
| **Valid** | 139 | 20.0 |
|
49 |
+
| **Test** | 140 | 20.1 |
|
50 |
+
| **Total** | 696 | 100.0 |
|
51 |
+
|
52 |
+
|
53 |
+
|
54 |
+
## Visualization
|
55 |
+
|
56 |
+
You can visualize word‑cloud statistics, clip‑duration distributions, and topic breakdowns using standard Python plotting tools.
|
57 |
+
|
58 |
+
## Citation
|
59 |
+
|
60 |
+
```bibtex
|
61 |
+
@article{luo2025omniresponse,
|
62 |
+
title={OmniResponse: Online Multimodal Conversational Response Generation in Dyadic Interactions},
|
63 |
+
author={Luo, Cheng and Wang, Jianghui and Li, Bing and Song, Siyang and Ghanem, Bernard},
|
64 |
+
journal={arXiv preprint arXiv:2505.21724},
|
65 |
+
year={2025}
|
66 |
+
}}
|
67 |
+
```
|
68 |
+
|
69 |
+
## License
|
70 |
+
|
71 |
+
This dataset is released under the **CC BY-NC 4.0** license.
|
72 |
+
|
73 |
+
```
|
74 |
+
```
|