zhoubingyu/KABR
收藏Hugging Face2026-04-20 更新2026-04-26 收录
下载链接：
https://hf-mirror.com/datasets/zhoubingyu/KABR
下载链接
链接失效反馈官方服务：
资源简介：
---
license: cc0-1.0
task_categories:
- video-classification
tags:
- zebra
- giraffe
- plains zebra
- Grevy's zebra
- video
- animal behavior
- behavior recognition
- annotation
- annotated video
- conservation
- drone
- UAV
- imbalanced
- Kenya
- Mpala Research Centre
pretty_name: >-
KABR: In-Situ Dataset for Kenyan Animal Behavior Recognition from Drone
Videos
description: "Initial KABR project release, contains drone video clips (mini-scenes) of giraffes, plains zebras, and Grevy's zebras with behavior labels from a subset of videos collected at the Mpala Research Centre in January 2023."
size_categories:
- 1M<n<10M
---
# Dataset Card for KABR: In-Situ Dataset for Kenyan Animal Behavior Recognition from Drone Videos
## Dataset Description
- **Homepage:** [KABR Mini-Scene Site](https://kabrdata.xyz/)
- **Project Page:** [KABR Site](https://imageomics.github.io/KABR/)
- **Repository:** https://github.com/Imageomics/kabr-tools
- **Paper:** https://openaccess.thecvf.com/content/WACV2024W/CV4Smalls/papers/Kholiavchenko_KABR_In-Situ_Dataset_for_Kenyan_Animal_Behavior_Recognition_From_Drone_WACVW_2024_paper.pdf
### Dataset Summary
We present a novel high-quality dataset for animal behavior recognition from drone videos.
The dataset is focused on Kenyan wildlife and contains behaviors of giraffes, plains zebras, and Grevy's zebras.
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.
In the annotation process for this dataset, a team of 10 people was involved, with an expert zoologist overseeing the process.
Each behavior was labeled based on its distinctive features, using a standardized set of criteria to ensure consistency and accuracy across the annotations.
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.
The drone footage is captured at a resolution of 5472 x 3078 pixels, and the videos were recorded at a frame rate of 29.97 frames per second.
<!--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).-->
### Supported Tasks and Leaderboards
The results of our evaluation using I3D, SlowFast, and X3D architectures are given in the table below. For each one, the model was trained for 120 epochs with batch size of 5. For more information on these results, see our [paper](coming soon).
| Method | All | Giraffes | Plains Zebras | Grevy’s Zebras |
| ---- | ---- | ---- | ---- | ---- |
| I3D (16x5) | 53.41 | 61.82 | 58.75 | 46.73 |
| SlowFast (16x5, 4x5) | 52.92 | 61.15 | 60.60 | 47.42 |
| X3D (16x5) | 61.9 | 65.1 | 63.11 | 51.16 |
### Languages
English
## Dataset Structure
Under `KABR/dataset/image/`, the data has been archived into `.zip` files, which are split into 2GB files. These must be recombined and extracted.
After cloning and navigating into the repository, you can use the following commands to do the reconstruction:
```bash
cd KABR/dataset/image/
cat giraffes_part_* > giraffes.zip
md5sum giraffes.zip # Compare this to what's shown with `cat giraffes_md5.txt`
unzip giraffes.zip
rm -rf giraffes_part_*
# Similarly for `zebras_grevys_part_*` and `zebras_plains_part_*`
```
Alternatively, there is a download script, `download.py`, which allows a download of the entire dataset in its established format without requiring one to clone the repository (cloning requires _at least_ double the size of the dataset to store). To proceed with this approach, download `download.py` to the system where you want to access the data.
Then, in the same directory as the script, run the following to begin the download:
```
pip install requests
python download.py
```
This script then downloads all the files present in the repository (without making a clone of the `.git` directory, etc.), concatenates the part files to their ZIP archives, verifies the MD5 checksums, extracts, and cleans up so that the folder structure, as described below, is present.
Note that it will require approximately 116GB of free space to complete this process, though the final dataset will only take about 61GB of disk space (the script removes the extra files after checking the download was successful).
The KABR dataset follows the Charades format:
```
KABR
/dataset
/image
/video_1
/image_1.jpg
/image_2.jpg
...
/image_n.jpg
/video_2
/image_1.jpg
/image_2.jpg
...
/image_n.jpg
...
/video_n
/image_1.jpg
/image_2.jpg
/image_3.jpg
...
/image_n.jpg
/annotation
/classes.json
/train.csv
/val.csv
```
The dataset can be directly loaded and processed by the [SlowFast](https://github.com/facebookresearch/SlowFast) framework.
**Informational Files**
* `KABR/configs`: examples of SlowFast framework configs.
* `KABR/annotation/distribution.xlsx`: distribution of classes for all videos.
**Scripts:**
* `image2video.py`: Encode image sequences into the original video.
* For example, `[image/G0067.1, image/G0067.2, ..., image/G0067.24]` will be encoded into `video/G0067.mp4`.
* `image2visual.py`: Encode image sequences into the original video with corresponding annotations.
* For example, `[image/G0067.1, image/G0067.2, ..., image/G0067.24]` will be encoded into `visual/G0067.mp4`.
### Data Instances
**Naming:** Within the image folder, the `video_n` folders are named as follows (X indicates a number):
* G0XXX.X - Giraffes
* ZP0XXX.X - Plains Zebras
* ZG0XXX.X - Grevy's Zebras
* Within each of these folders the images are simply `X.jpg`.
**Note:** The dataset consists of a total of 1,139,893 frames captured from drone videos. There are 488,638 frames of Grevy's zebras, 492,507 frames of plains zebras, and 158,748 frames of giraffes.
### Data Fields
There are 14,764 unique behavioral sequences in the dataset. These consist of eight distinct behaviors:
- Walk
- Trot
- Run: animal is moving at a cantor or gallop
- Graze: animal is eating grass or other vegetation
- Browse: animal is eating trees or bushes
- Head Up: animal is looking around or observe surroundings
- Auto-Groom: animal is grooming itself (licking, scratching, or rubbing)
- Occluded: animal is not fully visible
### Data Splits
Training and validation sets are indicated by their respective CSV files (`train.csv` and `val.csv`), located within the `annotation` folder.
## Dataset Creation
### Curation Rationale
We present a novel high-quality dataset for animal behavior recognition from drone videos.
The dataset is focused on Kenyan wildlife and contains behaviors of giraffes, plains zebras, and Grevy's zebras.
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.
In the annotation process for this dataset, a team of 10 people was involved, with an expert zoologist overseeing the process.
Each behavior was labeled based on its distinctive features, using a standardized set of criteria to ensure consistency and accuracy across the annotations.
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.
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.
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.
<!-- [To be added:] -->
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.
The results demonstrate the effectiveness of the dataset for animal behavior recognition and highlight the potential for further research in this area.
### Source Data
#### Initial Data Collection and Normalization
Data was collected from 6 January 2023 through 21 January 2023 at the [Mpala Research Centre](https://mpala.org/) in Kenya under a Nacosti research license. We used DJI Mavic 2S drones equipped with cameras to record 5.4K resolution videos (5472 x 3078 pixels) from varying altitudes and distances of 10 to 50 meters from the animals (distance was determined by circumstances and safety regulations).
Mini-scenes were extracted from these videos to reduce the impact of drone movement and facilitate human annotation. Animals were detected in frame using YOLOv8, then the SORT tracking algorithm was applied to follow their movement. A 400 by 300 pixel window, centered on the animal, was then extracted; this is the mini-scene.
<!--
#### Who are the source language producers?
[More Information Needed]
-->
### Annotations
#### Annotation process
In the annotation process for this dataset, a team of 10 people was involved, with an expert zoologist overseeing the process.
Each behavior was labeled based on its distinctive features, using a standardized set of criteria to ensure consistency and accuracy across the annotations.
<!--
#### Who are the annotators?
[More Information Needed]
-->
### Personal and Sensitive Information
Though there are endangered species included in this data, exact locations are not provided and their safety is assured by their location within the preserve.
## Considerations for Using the Data
<!--
### Social Impact of Dataset
[More Information Needed]
### Discussion of Biases
[More Information Needed]
-->
### Other Known Limitations
This data exhibits a long-tailed distribution due to the natural variation in frequency of the observed behaviors.
## Additional Information
### Authors
* Maksim Kholiavchenko (Rensselaer Polytechnic Institute) - ORCID: 0000-0001-6757-1957
* Jenna Kline (The Ohio State University) - ORCID: 0009-0006-7301-5774
* Michelle Ramirez (The Ohio State University)
* Sam Stevens (The Ohio State University)
* Alec Sheets (The Ohio State University) - ORCID: 0000-0002-3737-1484
* Reshma Ramesh Babu (The Ohio State University) - ORCID: 0000-0002-2517-5347
* Namrata Banerji (The Ohio State University) - ORCID: 0000-0001-6813-0010
* Elizabeth Campolongo (Imageomics Institute, The Ohio State University) - ORCID: 0000-0003-0846-2413
* Matthew Thompson (Imageomics Institute, The Ohio State University) - ORCID: 0000-0003-0583-8585
* Nina Van Tiel (Eidgenössische Technische Hochschule Zürich) - ORCID: 0000-0001-6393-5629
* Jackson Miliko (Mpala Research Centre)
* Eduardo Bessa (Universidade de Brasília) - ORCID: 0000-0003-0606-5860
* Tanya Berger-Wolf (The Ohio State University) - ORCID: 0000-0001-7610-1412
* Daniel Rubenstein (Princeton University) - ORCID: 0000-0001-9049-5219
* Charles Stewart (Rensselaer Polytechnic Institute)
### Licensing Information
This dataset is dedicated to the public domain for the benefit of scientific pursuits. We ask that you cite the dataset and journal paper using the below citations if you make use of it in your research.
### Citation Information
#### Dataset
```
@misc{KABR_Data,
author = {Kholiavchenko, Maksim and Kline, Jenna and Ramirez, Michelle and Stevens, Sam and Sheets, Alec and Babu, Reshma and Banerji, Namrata and Campolongo, Elizabeth and Thompson, Matthew and Van Tiel, Nina and Miliko, Jackson and Bessa, Eduardo and Duporge, Isla and Berger-Wolf, Tanya and Rubenstein, Daniel and Stewart, Charles},
title = {KABR: In-Situ Dataset for Kenyan Animal Behavior Recognition from Drone Videos},
year = {2023},
url = {https://huggingface.co/datasets/imageomics/KABR},
doi = {10.57967/hf/1010},
publisher = {Hugging Face}
}
```
#### Paper
```
@inproceedings{kholiavchenko2024kabr,
title={KABR: In-Situ Dataset for Kenyan Animal Behavior Recognition from Drone Videos},
author={Kholiavchenko, Maksim and Kline, Jenna and Ramirez, Michelle and Stevens, Sam and Sheets, Alec and Babu, Reshma and Banerji, Namrata and Campolongo, Elizabeth and Thompson, Matthew and Van Tiel, Nina and Miliko, Jackson and Bessa, Eduardo and Duporge, Isla and Berger-Wolf, Tanya and Rubenstein, Daniel and Stewart, Charles},
booktitle={Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision},
pages={31-40},
year={2024}
}
```
### Contributions
This work was supported by the [Imageomics Institute](https://imageomics.org), which is funded by the US National Science Foundation's Harnessing the Data Revolution (HDR) program under [Award #2118240](https://www.nsf.gov/awardsearch/showAward?AWD_ID=2118240) (Imageomics: A New Frontier of Biological Information Powered by Knowledge-Guided Machine Learning). Additional support was also provided by the [AI Institute for Intelligent Cyberinfrastructure with Computational Learning in the Environment (ICICLE)](https://icicle.osu.edu/), which is funded by the US National Science Foundation under [Award #2112606](https://www.nsf.gov/awardsearch/showAward?AWD_ID=2112606). Any opinions, findings and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.
The data was gathered at the [Mpala Research Centre](https://mpala.org/) in Kenya, in accordance with Research License No. NACOSTI/P/22/18214. The data collection protocol adhered strictly to the guidelines set forth by the Institutional Animal Care and Use Committee under permission No. IACUC 1835F.
许可证：CC0 1.0（cc0-1.0）
任务类别：
- 视频分类（video-classification）
标签：
- 斑马（zebra）
- 长颈鹿（giraffe）
- 平原斑马（plains zebra）
- 细纹斑马（Grevy's zebra）
- 视频（video）
- 动物行为（animal behavior）
- 行为识别（behavior recognition）
- 标注（annotation）
- 标注视频（annotated video）
- 保护（conservation）
- 无人机（drone）
- 无人飞行器（UAV）
- 类别不平衡（imbalanced）
- 肯尼亚（Kenya）
- 姆帕拉研究中心（Mpala Research Centre）
易读名称：KABR：基于无人机视频的肯尼亚动物行为识别原位数据集（KABR: In-Situ Dataset for Kenyan Animal Behavior Recognition from Drone Videos）
描述：本数据集为KABR项目的初始发布版本，包含2023年1月于肯尼亚姆帕拉研究中心采集的部分无人机视频片段（迷你场景），涵盖长颈鹿、平原斑马及细纹斑马，并为其中部分视频标注了行为标签。
尺寸类别：100万<n<1000万帧（1M<n<10M）
---
# KABR数据集卡片：基于无人机视频的肯尼亚动物行为识别原位数据集
## 数据集概述
- **主页**：[KABR迷你场景站点](https://kabrdata.xyz/)
- **项目页面**：[KABR站点](https://imageomics.github.io/KABR/)
- **代码仓库**：https://github.com/Imageomics/kabr-tools
- **论文**：https://openaccess.thecvf.com/content/WACV2024W/CV4Smalls/papers/Kholiavchenko_KABR_In-Situ_Dataset_for_Kenyan_Animal_Behavior_Recognition_From_Drone_WACVW_2024_paper.pdf
### 数据集摘要
本数据集面向无人机视频中的动物行为识别任务，提供了一套全新的高质量标注数据。数据集聚焦肯尼亚野生动物，涵盖长颈鹿、平原斑马及细纹斑马的行为数据。本数据集包含超过10小时的标注视频，共包含8个类别：7种动物行为类别，以及1个遮挡实例专用类别。
本数据集的标注工作由10人组成的团队完成，并由一名动物学专家全程监督。所有行为标签均基于动物的独特特征，采用标准化的标注准则进行标注，以确保全数据集标注的一致性与准确性。
数据集采用无人机在肯尼亚姆帕拉研究中心（Mpala Research Centre）上空对野生动物进行拍摄，获取了动物自然行为的高质量视频素材。无人机拍摄的视频分辨率为5472×3078像素，帧率为29.97帧每秒。
### 支持任务与评测榜单
我们采用I3D、SlowFast及X3D三种架构开展模型评测，评测结果如下表所示。所有模型均以批量大小5训练120个epoch。如需了解更多评测细节，请参阅我们的[论文（即将上线）](coming soon)。
| 方法 | 全部 | 长颈鹿 | 平原斑马 | 细纹斑马 |
| ---- | ---- | ---- | ---- | ---- |
| I3D (16x5) | 53.41 | 61.82 | 58.75 | 46.73 |
| SlowFast (16x5, 4x5) | 52.92 | 61.15 | 60.60 | 47.42 |
| X3D (16x5) | 61.9 | 65.1 | 63.11 | 51.16 |
### 语言
英语
## 数据集结构
数据集文件归档于`KABR/dataset/image/`路径下，以2GB为单位拆分为多个分卷压缩包，需先合并解压后方可使用。克隆代码仓库并进入对应目录后，可通过以下命令完成分卷合并与解压：
bash
cd KABR/dataset/image/
cat giraffes_part_* > giraffes.zip
md5sum giraffes.zip # Compare this to what's shown with `cat giraffes_md5.txt`
unzip giraffes.zip
rm -rf giraffes_part_*
# Similarly for `zebras_grevys_part_*` and `zebras_plains_part_*`
此外，本数据集还提供了下载脚本`download.py`，无需克隆完整代码仓库即可直接下载并自动处理全量数据集（克隆仓库所需的存储空间至少为数据集本身的两倍）。使用该脚本的步骤如下：
1. 将`download.py`下载至目标存储设备
2. 在脚本所在目录执行以下命令开始下载：
pip install requests
python download.py
该脚本将自动下载仓库中的所有文件（不会克隆.git目录等额外内容），合并分卷压缩包，验证MD5校验和，完成解压并清理临时文件，最终生成如下所述的标准文件夹结构。
整个下载及处理过程需占用约116GB的临时存储空间，最终数据集仅需约61GB的磁盘空间（脚本将在验证下载成功后自动删除临时文件）。
本KABR数据集遵循查德斯格式（Charades），文件夹结构如下：
KABR
/dataset
/image
/video_1
/image_1.jpg
/image_2.jpg
...
/image_n.jpg
/video_2
/image_1.jpg
/image_2.jpg
...
/image_n.jpg
...
/video_n
/image_1.jpg
/image_2.jpg
/image_3.jpg
...
/image_n.jpg
/annotation
/classes.json
/train.csv
/val.csv
本数据集可直接通过SlowFast框架加载并处理。
**辅助文件**
* `KABR/configs`：SlowFast框架的配置示例文件
* `KABR/annotation/distribution.xlsx`：全视频的类别分布统计文件
**辅助脚本**
* `image2video.py`：将图像序列编码为原始视频。例如，将`[image/G0067.1, image/G0067.2, ..., image/G0067.24]`编码为`video/G0067.mp4`。
* `image2visual.py`：将图像序列编码为带有对应标注的可视化视频。例如，将`[image/G0067.1, image/G0067.2, ..., image/G0067.24]`编码为`visual/G0067.mp4`。
### 数据实例
**命名规则**：图像文件夹下的`video_n`文件夹命名规则如下（X代表数字）：
* G0XXX.X - 长颈鹿类视频
* ZP0XXX.X - 平原斑马类视频
* ZG0XXX.X - 细纹斑马类视频
每个文件夹内的图像均以`X.jpg`命名。
**说明**：本数据集共包含1,139,893帧无人机拍摄的视频帧，其中细纹斑马帧488,638帧，平原斑马帧492,507帧，长颈鹿帧158,748帧。
### 数据字段
本数据集共包含14,764个独立的行为序列，涵盖8种不同的行为类别：
- 行走（Walk）
- 小跑（Trot）
- 奔跑（Run：动物以慢步快跑或疾驰的方式移动）
- 啃食草本植被（Graze：动物食用青草或其他草本植物）
- 啃食木本植被（Browse：动物食用树木或灌丛）
- 抬头观察（Head Up：动物环顾四周或观察周遭环境）
- 自我理毛（Auto-Groom：动物通过舔舐、抓挠或摩擦进行自我梳理）
- 遮挡（Occluded：动物未完全可见）
### 数据划分
训练集与验证集分别由`annotation`文件夹下的`train.csv`和`val.csv`文件指定。
## 数据集构建
### 数据集构建初衷
本数据集面向无人机视频中的动物行为识别任务，提供了一套全新的高质量标注数据。数据集聚焦肯尼亚野生动物，涵盖长颈鹿、平原斑马及细纹斑马的行为数据。本数据集包含超过10小时的标注视频，共包含8个类别：7种动物行为类别，以及1个遮挡实例专用类别。
本数据集的标注工作由10人组成的团队完成，并由一名动物学专家全程监督。所有行为标签均基于动物的独特特征，采用标准化的标注准则进行标注，以确保全数据集标注的一致性与准确性。
数据集采用无人机在肯尼亚姆帕拉研究中心上空对野生动物进行拍摄，获取了动物自然行为的高质量视频素材。
我们认为本数据集将为从事动物行为识别研究的科研人员提供宝贵的资源：其包含的多样化高质量标注视频可用于评估深度学习模型，同时也可用于研究肯尼亚野生动物的行为模式，为野生动物保护与管理策略提供数据支撑。
此外，本数据集还提供了详细的数据集说明与标注流程，并基于传统深度学习模型开展了初步的实验验证。实验结果证明了本数据集在动物行为识别任务中的有效性，同时也凸显了该领域后续研究的潜力。
### 源数据
#### 初始数据采集与标准化
数据采集于2023年1月6日至2023年1月21日，地点为肯尼亚姆帕拉研究中心（Mpala Research Centre），并获得了NACOSTI科研许可。本次采集采用DJI Mavic 2S无人机搭载相机，以5.4K分辨率（5472×3078像素）录制视频，飞行高度与动物的距离为10至50米（距离由实际场景与安全规范决定）。
我们从原始视频中提取迷你场景，以降低无人机运动对标注的影响并简化人工标注流程。首先使用YOLOv8在帧中检测动物，随后采用SORT跟踪算法追踪动物的运动轨迹，最终提取以动物为中心的400×300像素窗口作为迷你场景。
### 标注信息
#### 标注流程
本数据集的标注工作由10人组成的团队完成，并由一名动物学专家全程监督。所有行为标签均基于动物的独特特征，采用标准化的标注准则进行标注，以确保全数据集标注的一致性与准确性。
### 个人与敏感信息
尽管本数据集包含濒危物种，但未提供其精确位置，且所有拍摄区域均位于保护区内，可确保动物的安全。
## 数据使用注意事项
### 其他已知局限性
由于观测到的行为频率存在自然差异，本数据集呈现长尾分布特征。
## 附加信息
### 作者
* Maksim Kholiavchenko（伦斯勒理工学院） - ORCID: 0000-0001-6757-1957
* Jenna Kline（俄亥俄州立大学） - ORCID: 0009-0006-7301-5774
* Michelle Ramirez（俄亥俄州立大学）
* Sam Stevens（俄亥俄州立大学）
* Alec Sheets（俄亥俄州立大学） - ORCID: 0000-0002-3737-1484
* Reshma Ramesh Babu（俄亥俄州立大学） - ORCID: 0000-0002-2517-5347
* Namrata Banerji（俄亥俄州立大学） - ORCID: 0000-0001-6813-0010
* Elizabeth Campolongo（Imageomics研究所，俄亥俄州立大学） - ORCID: 0000-0003-0846-2413
* Matthew Thompson（Imageomics研究所，俄亥俄州立大学） - ORCID: 0000-0003-0583-8585
* Nina Van Tiel（苏黎世联邦理工学院） - ORCID: 0000-0001-6393-5629
* Jackson Miliko（姆帕拉研究中心）
* Eduardo Bessa（巴西利亚大学） - ORCID: 0000-0003-0606-5860
* Tanya Berger-Wolf（俄亥俄州立大学） - ORCID: 0000-0001-7610-1412
* Daniel Rubenstein（普林斯顿大学） - ORCID: 0000-0001-9049-5219
* Charles Stewart（伦斯勒理工学院）
### 许可信息
本数据集已捐赠至公共领域，旨在促进科学研究。如您在研究中使用本数据集，请引用以下格式的数据集与论文文献。
### 引用信息
#### 数据集引用
@misc{KABR_Data,
author = {Kholiavchenko, Maksim and Kline, Jenna and Ramirez, Michelle and Stevens, Sam and Sheets, Alec and Babu, Reshma and Banerji, Namrata and Campolongo, Elizabeth and Thompson, Matthew and Van Tiel, Nina and Miliko, Jackson and Bessa, Eduardo and Duporge, Isla and Berger-Wolf, Tanya and Rubenstein, Daniel and Stewart, Charles},
title = {KABR: In-Situ Dataset for Kenyan Animal Behavior Recognition from Drone Videos},
year = {2023},
url = {https://huggingface.co/datasets/imageomics/KABR},
doi = {10.57967/hf/1010},
publisher = {Hugging Face}
}
#### 论文引用
@inproceedings{kholiavchenko2024kabr,
title={KABR: In-Situ Dataset for Kenyan Animal Behavior Recognition from Drone Videos},
author={Kholiavchenko, Maksim and Kline, Jenna and Ramirez, Michelle and Stevens, Sam and Sheets, Alec and Babu, Reshma and Banerji, Namrata and Campolongo, Elizabeth and Thompson, Matthew and Van Tiel, Nina and Miliko, Jackson and Bessa, Eduardo and Duporge, Isla and Berger-Wolf, Tanya and Rubenstein, Daniel and Stewart, Charles},
booktitle={Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision},
pages={31-40},
year={2024}
}
### 资助信息
本研究由Imageomics研究所资助，该研究所由美国国家科学基金会（NSF）数据革新计划（Harnessing the Data Revolution, HDR）资助，项目编号为#2118240（Imageomics：知识引导机器学习驱动的生物信息学新前沿）。此外，本研究还得到了智能网络基础设施人工智能研究所（ICICLE）的资助，该研究所由美国国家科学基金会资助，项目编号为#2112606。本材料中的任何观点、发现、结论或建议均为作者个人观点，不一定代表美国国家科学基金会的官方立场。
本数据集采集于肯尼亚姆帕拉研究中心（Mpala Research Centre），并获得了科研许可编号NACOSTI/P/22/18214。数据采集流程严格遵循机构动物护理与使用委员会的指导方针，许可编号为IACUC 1835F。
提供机构：
zhoubingyu
