SAE-AAI/FLIR_IR_Expansion

---
dataset_info:
features:
- name: image
dtype: image
- name: bounding_boxes
sequence:
sequence: float64
- name: classes
sequence: int64
splits:
- name: train
num_bytes: 147837671.888
num_examples: 2596
- name: validation
num_bytes: 26070707
num_examples: 458
download_size: 172845541
dataset_size: 173908378.888
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*
license: apache-2.0
task_categories:
- image-classification
pretty_name: FLIR IR YOLO expansion
size_categories:
- 1K<n<10K
---

# FLIR IR YOLO expansion

Images from repository [FLIR_IR_Expansion](https://github.com/sensationTI/FLIR_IR_Expansion)

This expansion pack is prepared specifically for training a YOU-ONLY-LOOK-ONCE(YOLO) network. All frames are labeled in the YOLO format. If you want to use this expansion pack for other purposes, images are still available for download but requires manual labeling

## Download all dataset

If you want to download all dataset you must do

```
from datasets import load_dataset

dataset = load_dataset("SAE-AAI/FLIR_IR_Expansion")
```

If you only want to download `train` or `validation` split you must do

```
from datasets import load_dataset

train_dataset = load_dataset("SAE-AAI/FLIR_IR_Expansion", split='train')
validation_dataset = load_dataset("SAE-AAI/FLIR_IR_Expansion", split='validation')
```

## Download by stream

If you want to donwload dataset by stream, you must do

```
from datasets import load_dataset

iterable_dataset = load_dataset("SAE-AAI/FLIR_IR_Expansion", streaming=True)
```

now you can get every sample by

```
for sample in iterable_dataset['train']:
print(sample['bounding_boxes'], sample['classes'])
example['sample'].show()
break
```

or with

```
sample = next(iter(iterable_dataset['train']))
print(sample['bounding_boxes'], sample['classes'])
sample['image']
```

If you want to get a batch of samples you ca do

```
BS = 4
example_batch = list(iterable_dataset['train'].take(BS))
```

原始信息汇总

数据集概述

数据集名称

• FLIR IR YOLO expansion

数据集特征

• image: 图像数据
• bounding_boxes: 边界框数据，序列类型为float64
• classes: 类别数据，序列类型为int64

数据集分割

• train: 2596个样本，占用空间147837671.888字节
• validation: 458个样本，占用空间26070707字节

数据集大小

• 下载大小: 172845541字节
• 数据集总大小: 173908378.888字节

配置信息

• config_name: default
• data_files:
  • train: data/train-*
  • validation: data/validation-*

许可证

• apache-2.0

任务类别

• image-classification

数据集大小类别

• 1K<n<10K