Datasets:
Initializing README
Browse files
README.md
CHANGED
@@ -1,3 +1,205 @@
|
|
1 |
---
|
2 |
license: cc0-1.0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: cc0-1.0
|
3 |
+
task_categories:
|
4 |
+
- video-classification
|
5 |
+
tags:
|
6 |
+
- zebra
|
7 |
+
- giraffe
|
8 |
+
- plains zebra
|
9 |
+
- Grevy's zebra
|
10 |
+
- video
|
11 |
+
- animal behavior
|
12 |
+
- behavior recognition
|
13 |
+
- annotation
|
14 |
+
- annotated video
|
15 |
+
- conservation
|
16 |
+
- drone
|
17 |
+
- UAV
|
18 |
+
- imbalanced
|
19 |
+
- Kenya
|
20 |
+
- Mpala Research Centre
|
21 |
+
pretty_name: >-
|
22 |
+
KABR: High-Quality Dataset for Kenyan Animal Behavior Recognition from Drone
|
23 |
+
Videos
|
24 |
+
size_categories:
|
25 |
+
- 1M<n<10M
|
26 |
---
|
27 |
+
# Dataset Card for KABR: High-Quality Dataset for Kenyan Animal Behavior Recognition from Drone Videos
|
28 |
+
|
29 |
+
## Dataset Description
|
30 |
+
|
31 |
+
- **Homepage:** https://dirtmaxim.github.io/kabr/
|
32 |
+
- **Repository:** https://github.com/dirtmaxim/kabr-tools
|
33 |
+
- **Paper:** [Coming Soon]
|
34 |
+
- **Leaderboard:**
|
35 |
+
- **Point of Contact:**
|
36 |
+
|
37 |
+
### Dataset Summary
|
38 |
+
|
39 |
+
We present a novel high-quality dataset for animal behavior recognition from drone videos.
|
40 |
+
The dataset is focused on Kenyan wildlife and contains behaviors of giraffes, plains zebras, and Grevy's zebras.
|
41 |
+
The dataset consists of more than 10 hours of annotated videos, and it includes eight different classes, encompassing seven types of animal behavior and an additional category for occluded instances.
|
42 |
+
In the annotation process for this dataset, a team of 10 people was involved, with an expert zoologist overseeing the process.
|
43 |
+
Each behavior was labeled based on its distinctive features, using a standardized set of criteria to ensure consistency and accuracy across the annotations.
|
44 |
+
The dataset was collected using drones that flew over the animals in the [Mpala Research Centre](https://mpala.org/) in Kenya, providing high-quality video footage of the animal's natural behaviors.
|
45 |
+
|
46 |
+
<!--This dataset card aims to be a base template for new datasets. It has been generated using [this raw template](https://github.com/huggingface/huggingface_hub/blob/main/src/huggingface_hub/templates/datasetcard_template.md?plain=1).-->
|
47 |
+
|
48 |
+
### Supported Tasks and Leaderboards
|
49 |
+
|
50 |
+
[Include Benchmarks Here]
|
51 |
+
|
52 |
+
### Languages
|
53 |
+
|
54 |
+
[More Information Needed]
|
55 |
+
|
56 |
+
## Dataset Structure
|
57 |
+
|
58 |
+
The KABR dataset follows the Charades format:
|
59 |
+
|
60 |
+
```
|
61 |
+
KABR
|
62 |
+
/images
|
63 |
+
/video_1
|
64 |
+
/image_1.jpg
|
65 |
+
/image_2.jpg
|
66 |
+
...
|
67 |
+
/image_n.jpg
|
68 |
+
/video_2
|
69 |
+
/image_1.jpg
|
70 |
+
/image_2.jpg
|
71 |
+
...
|
72 |
+
/image_n.jpg
|
73 |
+
...
|
74 |
+
/video_n
|
75 |
+
/image_1.jpg
|
76 |
+
/image_2.jpg
|
77 |
+
/image_3.jpg
|
78 |
+
...
|
79 |
+
/image_n.jpg
|
80 |
+
/annotation
|
81 |
+
/classes.json
|
82 |
+
/train.csv
|
83 |
+
/val.csv
|
84 |
+
```
|
85 |
+
|
86 |
+
The dataset can be directly loaded and processed by the [SlowFast](https://github.com/facebookresearch/SlowFast) framework.
|
87 |
+
|
88 |
+
**Informational Files**
|
89 |
+
* `KABR/configs`: examples of SlowFast framework configs.
|
90 |
+
* `KABR/annotation/distribution.xlsx`: distribution of classes for all videos.
|
91 |
+
|
92 |
+
**Scripts:**
|
93 |
+
* `image2video.py`: Encode image sequences into the original video.
|
94 |
+
* For example, `[image/G0067.1, image/G0067.2, ..., image/G0067.24]` will be encoded into `video/G0067.mp4`.
|
95 |
+
* `image2visual.py`: Encode image sequences into the original video with corresponding annotations.
|
96 |
+
* For example, `[image/G0067.1, image/G0067.2, ..., image/G0067.24]` will be encoded into `visual/G0067.mp4`.
|
97 |
+
|
98 |
+
### Data Instances
|
99 |
+
|
100 |
+
[More Information Needed]
|
101 |
+
**Naming:** Within the image folder, the `video_n` folders are named as follows (X indicates a number):
|
102 |
+
* G0XXX.X - Giraffes
|
103 |
+
* ZP0XXX.X - Plains Zebras
|
104 |
+
* ZG0XXX.X - Grevy's Zebras
|
105 |
+
* Within each of these folders the images are simply `X.jpg`.
|
106 |
+
|
107 |
+
### Data Fields
|
108 |
+
|
109 |
+
[More Information Needed]
|
110 |
+
|
111 |
+
### Data Splits
|
112 |
+
|
113 |
+
Training and validation sets are indicated by their respecive CSV files, located within the `annotation` folder.
|
114 |
+
|
115 |
+
## Dataset Creation
|
116 |
+
|
117 |
+
### Curation Rationale
|
118 |
+
|
119 |
+
We present a novel high-quality dataset for animal behavior recognition from drone videos.
|
120 |
+
The dataset is focused on Kenyan wildlife and contains behaviors of giraffes, plains zebras, and Grevy's zebras.
|
121 |
+
The dataset consists of more than 10 hours of annotated videos, and it includes eight different classes, encompassing seven types of animal behavior and an additional category for occluded instances.
|
122 |
+
In the annotation process for this dataset, a team of 10 people was involved, with an expert zoologist overseeing the process.
|
123 |
+
Each behavior was labeled based on its distinctive features, using a standardized set of criteria to ensure consistency and accuracy across the annotations.
|
124 |
+
The dataset was collected using drones that flew over the animals in the [Mpala Research Centre](https://mpala.org/) in Kenya, providing high-quality video footage of the animal's natural behaviors.
|
125 |
+
We believe that this dataset will be a valuable resource for researchers working on animal behavior recognition, as it provides a diverse and high-quality set of annotated videos that can be used for evaluating deep learning models.
|
126 |
+
Additionally, the dataset can be used to study the behavior patterns of Kenyan animals and can help to inform conservation efforts and wildlife management strategies.
|
127 |
+
|
128 |
+
[To be added:]
|
129 |
+
We provide a detailed description of the dataset and its annotation process, along with some initial experiments on the dataset using conventional deep learning models.
|
130 |
+
The results demonstrate the effectiveness of the dataset for animal behavior recognition and highlight the potential for further research in this area.
|
131 |
+
|
132 |
+
### Source Data
|
133 |
+
|
134 |
+
#### Initial Data Collection and Normalization
|
135 |
+
|
136 |
+
[More Information Needed]
|
137 |
+
|
138 |
+
Data was collected from 6 January 2023 through 21 January 2023 at the [Mpala Research Centre](https://mpala.org/) in Kenya.
|
139 |
+
|
140 |
+
#### Who are the source language producers?
|
141 |
+
|
142 |
+
[More Information Needed]
|
143 |
+
|
144 |
+
### Annotations
|
145 |
+
|
146 |
+
#### Annotation process
|
147 |
+
|
148 |
+
[More Information Needed]
|
149 |
+
|
150 |
+
In the annotation process for this dataset, a team of 10 people was involved, with an expert zoologist overseeing the process.
|
151 |
+
Each behavior was labeled based on its distinctive features, using a standardized set of criteria to ensure consistency and accuracy across the annotations.
|
152 |
+
|
153 |
+
#### Who are the annotators?
|
154 |
+
|
155 |
+
[More Information Needed]
|
156 |
+
|
157 |
+
### Personal and Sensitive Information
|
158 |
+
|
159 |
+
[More Information Needed]
|
160 |
+
|
161 |
+
## Considerations for Using the Data
|
162 |
+
|
163 |
+
### Social Impact of Dataset
|
164 |
+
|
165 |
+
[More Information Needed]
|
166 |
+
|
167 |
+
### Discussion of Biases
|
168 |
+
|
169 |
+
[More Information Needed]
|
170 |
+
|
171 |
+
### Other Known Limitations
|
172 |
+
|
173 |
+
[More Information Needed]
|
174 |
+
|
175 |
+
## Additional Information
|
176 |
+
|
177 |
+
### Dataset Curators
|
178 |
+
|
179 |
+
[More Information Needed]
|
180 |
+
|
181 |
+
### Licensing Information
|
182 |
+
|
183 |
+
[More Information Needed]
|
184 |
+
|
185 |
+
### Citation Information
|
186 |
+
|
187 |
+
[More Information Needed]
|
188 |
+
|
189 |
+
### Contributions
|
190 |
+
|
191 |
+
<!---location for authors instead of under curators?--->
|
192 |
+
* Maksim Kholiavchenko (Rensselaer Polytechnic Institute) - ORCID: 0000-0001-6757-1957
|
193 |
+
* Jenna Kline (The Ohio State University)
|
194 |
+
* Michelle Ramirez (The Ohio State University)
|
195 |
+
* Sam Stevens (The Ohio State University)
|
196 |
+
* Alec Sheets (The Ohio State University) - ORCID: 0000-0002-3737-1484
|
197 |
+
* Reshma Ramesh Babu (The Ohio State University) - ORCID: 0000-0002-2517-5347
|
198 |
+
* Namrata Banerji (The Ohio State University) - ORCID: 0000-0001-6813-0010
|
199 |
+
* Elizabeth Campolongo (Imageomics Institute) - ORCID: 0000-0003-0846-2413
|
200 |
+
* Nina Van Tiel (Eidgenössische Technische Hochschule Zürich) - ORCID: 0000-0001-6393-5629
|
201 |
+
* Jackson Miliko (Mpala Research Centre)
|
202 |
+
* Eduardo Bessa (Universidade de Brasília) - ORCID: 0000-0003-0606-5860
|
203 |
+
* Tanya Berger-Wolf (The Ohio State University) - ORCID: 0000-0001-7610-1412
|
204 |
+
* Daniel Rubenstein (Princeton University) - ORCID: 0000-0001-9049-5219
|
205 |
+
* Charles Stewart (Rensselaer Polytechnic Institute)
|