Upload dataset
Browse files
README.md
CHANGED
|
@@ -1,3 +1,71 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
# LVBench Dataset
|
| 2 |
|
| 3 |
This is a video question-answering dataset with subsets: cartoon, documentary, live, selfmedia, sport, and tv.
|
|
|
|
| 1 |
+
---
|
| 2 |
+
configs:
|
| 3 |
+
- config_name: default
|
| 4 |
+
data_files:
|
| 5 |
+
- split: cartoon
|
| 6 |
+
path: data/cartoon-*
|
| 7 |
+
- split: documentary
|
| 8 |
+
path: data/documentary-*
|
| 9 |
+
- split: live
|
| 10 |
+
path: data/live-*
|
| 11 |
+
- split: selfmedia
|
| 12 |
+
path: data/selfmedia-*
|
| 13 |
+
- split: sport
|
| 14 |
+
path: data/sport-*
|
| 15 |
+
- split: tv
|
| 16 |
+
path: data/tv-*
|
| 17 |
+
dataset_info:
|
| 18 |
+
features:
|
| 19 |
+
- name: uid
|
| 20 |
+
dtype: string
|
| 21 |
+
- name: question
|
| 22 |
+
dtype: string
|
| 23 |
+
- name: answer
|
| 24 |
+
dtype: string
|
| 25 |
+
- name: question_type
|
| 26 |
+
sequence: string
|
| 27 |
+
- name: time_reference
|
| 28 |
+
dtype: string
|
| 29 |
+
- name: video_path
|
| 30 |
+
dtype: string
|
| 31 |
+
- name: video_info
|
| 32 |
+
struct:
|
| 33 |
+
- name: duration_minutes
|
| 34 |
+
dtype: float64
|
| 35 |
+
- name: fps
|
| 36 |
+
dtype: float64
|
| 37 |
+
- name: resolution
|
| 38 |
+
struct:
|
| 39 |
+
- name: height
|
| 40 |
+
dtype: int64
|
| 41 |
+
- name: width
|
| 42 |
+
dtype: int64
|
| 43 |
+
- name: key
|
| 44 |
+
dtype: string
|
| 45 |
+
- name: type
|
| 46 |
+
dtype: string
|
| 47 |
+
splits:
|
| 48 |
+
- name: cartoon
|
| 49 |
+
num_bytes: 84481
|
| 50 |
+
num_examples: 283
|
| 51 |
+
- name: documentary
|
| 52 |
+
num_bytes: 61632
|
| 53 |
+
num_examples: 197
|
| 54 |
+
- name: live
|
| 55 |
+
num_bytes: 76239
|
| 56 |
+
num_examples: 262
|
| 57 |
+
- name: selfmedia
|
| 58 |
+
num_bytes: 94943
|
| 59 |
+
num_examples: 317
|
| 60 |
+
- name: sport
|
| 61 |
+
num_bytes: 82412
|
| 62 |
+
num_examples: 240
|
| 63 |
+
- name: tv
|
| 64 |
+
num_bytes: 77531
|
| 65 |
+
num_examples: 250
|
| 66 |
+
download_size: 211694
|
| 67 |
+
dataset_size: 477238
|
| 68 |
+
---
|
| 69 |
# LVBench Dataset
|
| 70 |
|
| 71 |
This is a video question-answering dataset with subsets: cartoon, documentary, live, selfmedia, sport, and tv.
|