Niche-Squad/COLO

---
dataset_info:
- config_name: 0_all
features:
- name: image
dtype: image
- name: width
dtype: int64
- name: height
dtype: int64
- name: n_cows
dtype: int64
- name: annotations
sequence:
- name: id
dtype: int64
- name: image_id
dtype: int64
- name: category_id
dtype: int64
- name: iscrowd
dtype: int64
- name: area
dtype: float64
- name: bbox
sequence: float64
length: 4
- name: segmentation
sequence:
sequence: int64
- name: image_id
dtype: int64
- name: filename
dtype: string
splits:
- name: train
num_bytes: 130320762
num_examples: 904
- name: test
num_bytes: 13928675
num_examples: 100
download_size: 143829012
dataset_size: 144249437
- config_name: 1_top
features:
- name: image
dtype: image
- name: width
dtype: int64
- name: height
dtype: int64
- name: n_cows
dtype: int64
- name: annotations
sequence:
- name: id
dtype: int64
- name: image_id
dtype: int64
- name: category_id
dtype: int64
- name: iscrowd
dtype: int64
- name: area
dtype: float64
- name: bbox
sequence: float64
length: 4
- name: segmentation
sequence:
sequence: int64
- name: image_id
dtype: int64
- name: filename
dtype: string
splits:
- name: daylight
num_bytes: 53998347
num_examples: 296
- name: indoorlight
num_bytes: 23086697
num_examples: 118
- name: infrared
num_bytes: 11752283
num_examples: 90
- name: train
num_bytes: 80432409
num_examples: 454
- name: test
num_bytes: 8404918
num_examples: 50
download_size: 177400440
dataset_size: 177674654
- config_name: 2_side
features:
- name: image
dtype: image
- name: width
dtype: int64
- name: height
dtype: int64
- name: n_cows
dtype: int64
- name: annotations
sequence:
- name: id
dtype: int64
- name: image_id
dtype: int64
- name: category_id
dtype: int64
- name: iscrowd
dtype: int64
- name: area
dtype: float64
- name: bbox
sequence: float64
length: 4
- name: segmentation
sequence:
sequence: int64
- name: image_id
dtype: int64
- name: filename
dtype: string
splits:
- name: daylight
num_bytes: 36621130
num_examples: 290
- name: indoorlight
num_bytes: 14910133
num_examples: 113
- name: infrared
num_bytes: 3880850
num_examples: 97
- name: train
num_bytes: 49888354
num_examples: 450
- name: test
num_bytes: 5523758
num_examples: 50
download_size: 110254324
dataset_size: 110824225
- config_name: 3_external
features:
- name: image
dtype: image
- name: width
dtype: int64
- name: height
dtype: int64
- name: n_cows
dtype: int64
- name: annotations
sequence:
- name: id
dtype: int64
- name: image_id
dtype: int64
- name: category_id
dtype: int64
- name: iscrowd
dtype: int64
- name: area
dtype: float64
- name: bbox
sequence: float64
length: 4
- name: segmentation
sequence:
sequence: int64
- name: image_id
dtype: int64
- name: filename
dtype: string
splits:
- name: train
num_bytes: 30382759
num_examples: 200
- name: test
num_bytes: 7430774
num_examples: 50
download_size: 37623678
dataset_size: 37813533
- config_name: a1_t2s
features:
- name: image
dtype: image
- name: width
dtype: int64
- name: height
dtype: int64
- name: n_cows
dtype: int64
- name: annotations
sequence:
- name: id
dtype: int64
- name: image_id
dtype: int64
- name: category_id
dtype: int64
- name: iscrowd
dtype: int64
- name: area
dtype: float64
- name: bbox
sequence: float64
length: 4
- name: segmentation
sequence:
sequence: int64
- name: image_id
dtype: int64
- name: filename
dtype: string
splits:
- name: train
num_bytes: 88837326
num_examples: 504
- name: test
num_bytes: 5523758
num_examples: 50
download_size: 94192043
dataset_size: 94361084
- config_name: a2_s2t
features:
- name: image
dtype: image
- name: width
dtype: int64
- name: height
dtype: int64
- name: n_cows
dtype: int64
- name: annotations
sequence:
- name: id
dtype: int64
- name: image_id
dtype: int64
- name: category_id
dtype: int64
- name: iscrowd
dtype: int64
- name: area
dtype: float64
- name: bbox
sequence: float64
length: 4
- name: segmentation
sequence:
sequence: int64
- name: image_id
dtype: int64
- name: filename
dtype: string
splits:
- name: train
num_bytes: 55412111
num_examples: 500
- name: test
num_bytes: 8404918
num_examples: 50
download_size: 63528042
dataset_size: 63817029
- config_name: b_light
features:
- name: image
dtype: image
- name: width
dtype: int64
- name: height
dtype: int64
- name: n_cows
dtype: int64
- name: annotations
sequence:
- name: id
dtype: int64
- name: image_id
dtype: int64
- name: category_id
dtype: int64
- name: iscrowd
dtype: int64
- name: area
dtype: float64
- name: bbox
sequence: float64
length: 4
- name: segmentation
sequence:
sequence: int64
- name: image_id
dtype: int64
- name: filename
dtype: string
splits:
- name: train
num_bytes: 76120383
num_examples: 500
- name: test
num_bytes: 6280763
num_examples: 50
download_size: 82127375
dataset_size: 82401146
- config_name: c_external
features:
- name: image
dtype: image
- name: width
dtype: int64
- name: height
dtype: int64
- name: n_cows
dtype: int64
- name: annotations
sequence:
- name: id
dtype: int64
- name: image_id
dtype: int64
- name: category_id
dtype: int64
- name: iscrowd
dtype: int64
- name: area
dtype: float64
- name: bbox
sequence: float64
length: 4
- name: segmentation
sequence:
sequence: int64
- name: image_id
dtype: int64
- name: filename
dtype: string
splits:
- name: train
num_bytes: 144104201.292
num_examples: 1004
- name: test
num_bytes: 7430774
num_examples: 50
download_size: 151218220
dataset_size: 151534975.292
configs:
- config_name: 0_all
data_files:
- split: train
path: 0_all/train-*
- split: test
path: 0_all/test-*
- config_name: 1_top
data_files:
- split: daylight
path: 1_top/daylight-*
- split: indoorlight
path: 1_top/indoorlight-*
- split: infrared
path: 1_top/infrared-*
- split: train
path: 1_top/train-*
- split: test
path: 1_top/test-*
- config_name: 2_side
data_files:
- split: daylight
path: 2_side/daylight-*
- split: indoorlight
path: 2_side/indoorlight-*
- split: infrared
path: 2_side/infrared-*
- split: train
path: 2_side/train-*
- split: test
path: 2_side/test-*
- config_name: 3_external
data_files:
- split: train
path: 3_external/train-*
- split: test
path: 3_external/test-*
- config_name: a1_t2s
data_files:
- split: train
path: a1_t2s/train-*
- split: test
path: a1_t2s/test-*
- config_name: a2_s2t
data_files:
- split: train
path: a2_s2t/train-*
- split: test
path: a2_s2t/test-*
- config_name: b_light
data_files:
- split: train
path: b_light/train-*
- split: test
path: b_light/test-*
- config_name: c_external
data_files:
- split: train
path: c_external/train-*
- split: test
path: c_external/test-*
license: mit
task_categories:
- object-detection
tags:
- biology
pretty_name: COLO
size_categories:
- 1K<n<10K
---

# COw LOcalization (COLO) Dataset

The COw LOcalization (COLO) dataset is designed to localize cows in various indoor environments using different lighting conditions and view angles. This dataset offers 1,254 images and 11,818 cow instances, serving as a benchmark for the precision livestock farming community.


## Dataset Configurations

<style>
table {
width: 50%;
margin-left: auto;
margin-right: auto;
}
</style>

| **Configuration** | **Training Split** | **Testing Split** |
|:------------------|:-------------------|:---------------------|
| _0_all_ | Top-View + Side-View | Top-View + Side-View|
| _1_top_ | Top-View | Top-View |
| _2_side_ | Side-View | Side-View |
| _3_external_ | External | External |
| _a1_t2s_ | Top-View | Side-View |
| _a2_s2t_ | Side-View | Top-View |
| _b_light_ | Daylight | Indoor + NIR |
| _c_external_ | Top-View + Side-View | External |

## Download the Dataset

To download the dataset, you need to have the required Python dependencies installed. You can install them using either of the following commands:

```sh
python -m pip install pyniche
```
or

```sh
pip install pyniche
```

Once the dependencies are installed, use the Python console to provide the download destination folder in the `root` parameter and specify the export data format in the `format` parameter:

```python
from pyniche.data.download import COLO

# Example: Download COLO in the YOLO format
COLO(
root="download/yolo", # Destination folder
format="yolo", # Data format
)

# Example: Download COLO in the COCO format
COLO(
root="download/coco", # Destination folder
format="coco", # Data format
)
```

## Citation

[The page of the arXiv article](https://arxiv.org/abs/2407.20372)

```bibtex
@misc{das2024model,
title={A Model Generalization Study in Localizing Indoor Cows with COw LOcalization (COLO) dataset},
author={Mautushi Das and Gonzalo Ferreira and C. P. James Chen},
year={2024},
eprint={2407.20372},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
```

or

Das, M., Ferreira, G., & Chen, C. P. J. (2024). A Model Generalization Study in Localizing Indoor Cows with COw LOcalization (COLO) dataset. arXiv preprint arXiv:2407.20372

---

原始信息汇总

数据集概述

数据集配置

• config_name: 0_all

  • features:
    • image: 图片数据
    • width: 图片宽度，数据类型为int64
    • height: 图片高度，数据类型为int64
    • n_cows: 牛的数量，数据类型为int64
    • annotations: 注释信息，包含多个子特征
      • id: 注释ID，数据类型为int64
      • image_id: 图片ID，数据类型为int64
      • category_id: 类别ID，数据类型为int64
      • iscrowd: 是否为群体，数据类型为int64
      • area: 区域大小，数据类型为float64
      • bbox: 边界框，数据类型为float64，长度为4
      • segmentation: 分割信息，数据类型为int64
    • image_id: 图片ID，数据类型为int64
    • filename: 文件名，数据类型为string
  • splits:
    • train: 训练集，904个样本，占用130320762字节
    • test: 测试集，100个样本，占用13928675字节
  • download_size: 143829012字节
  • dataset_size: 144249437字节

• config_name: 1_top

  • features: 同上
  • splits:
    • daylight: 日光下数据，296个样本，占用53998347字节
    • indoorlight: 室内光下数据，118个样本，占用23086697字节
    • infrared: 红外光下数据，90个样本，占用11752283字节
    • train: 训练集，454个样本，占用80432409字节
    • test: 测试集，50个样本，占用8404918字节
  • download_size: 177400440字节
  • dataset_size: 177674654字节

• config_name: 2_side

  • features: 同上
  • splits:
    • daylight: 日光下数据，290个样本，占用36621130字节
    • indoorlight: 室内光下数据，113个样本，占用14910133字节
    • infrared: 红外光下数据，97个样本，占用3880850字节
    • train: 训练集，450个样本，占用49888354字节
    • test: 测试集，50个样本，占用5523758字节
  • download_size: 110254324字节
  • dataset_size: 110824225字节

• config_name: 3_external

  • features: 同上
  • splits:
    • train: 训练集，200个样本，占用30382759字节
    • test: 测试集，50个样本，占用7430774字节
  • download_size: 37623678字节
  • dataset_size: 37813533字节

• config_name: a1_t2s

  • features: 同上
  • splits:
    • train: 训练集，504个样本，占用88837326字节
    • test: 测试集，50个样本，占用5523758字节
  • download_size: 94192043字节
  • dataset_size: 94361084字节

• config_name: a2_s2t

  • features: 同上
  • splits:
    • train: 训练集，500个样本，占用55412111字节
    • test: 测试集，50个样本，占用8404918字节
  • download_size: 63528042字节
  • dataset_size: 63817029字节

• config_name: b_light

  • features: 同上
  • splits:
    • train: 训练集，500个样本，占用76120383字节
    • test: 测试集，50个样本，占用6280763字节
  • download_size: 82127375字节
  • dataset_size: 82401146字节

• config_name: c_external

  • features: 同上
  • splits:
    • train: 训练集，1004个样本，占用144104201.292字节
    • test: 测试集，50个样本，占用7430774字节
  • download_size: 151218220字节
  • dataset_size: 151534975.292字节

数据集文件路径

• config_name: 0_all

  • data_files:
    • train: 0_all/train-*
    • test: 0_all/test-*

• config_name: 1_top

  • data_files:
    • daylight: 1_top/daylight-*
    • indoorlight: 1_top/indoorlight-*
    • infrared: 1_top/infrared-*
    • train: 1_top/train-*
    • test: 1_top/test-*

• config_name: 2_side

  • data_files:
    • daylight: 2_side/daylight-*
    • indoorlight: 2_side/indoorlight-*
    • infrared: 2_side/infrared-*
    • train: 2_side/train-*
    • test: 2_side/test-*

• config_name: 3_external

  • data_files:
    • train: 3_external/train-*
    • test: 3_external/test-*

• config_name: a1_t2s

  • data_files:
    • train: a1_t2s/train-*
    • test: a1_t2s/test-*

• config_name: a2_s2t

  • data_files:
    • train: a2_s2t/train-*
    • test: a2_s2t/test-*

• config_name: b_light

  • data_files:
    • train: b_light/train-*
    • test: b_light/test-*

• config_name: c_external

  • data_files:
    • train: c_external/train-*
    • test: c_external/test-*

数据集许可证

• license: mit

任务类别

• task_categories: object-detection

标签

• tags: biology

数据集名称

• pretty_name: COLO

数据集大小

• size_categories: 1K<n<10K