IGNF/FLAIR_1_osm_clip

---
size_categories:
- 10K<n<100K
dataset_info:
features:
- name: image
dtype: image
- name: segmentation
dtype: image
- name: metadata
struct:
- name: address
struct:
- name: ISO3166-2-lvl4
dtype: string
- name: ISO3166-2-lvl6
dtype: string
- name: ISO3166-2-lvl7
dtype: string
- name: aerialway
dtype: string
- name: aeroway
dtype: string
- name: allotments
dtype: string
- name: amenity
dtype: string
- name: bridge
dtype: string
- name: building
dtype: string
- name: city
dtype: string
- name: city_district
dtype: string
- name: club
dtype: string
- name: commercial
dtype: string
- name: country
dtype: string
- name: country_code
dtype: string
- name: county
dtype: string
- name: craft
dtype: string
- name: emergency
dtype: string
- name: farm
dtype: string
- name: farmyard
dtype: string
- name: hamlet
dtype: string
- name: healthcare
dtype: string
- name: highway
dtype: string
- name: historic
dtype: string
- name: house_number
dtype: string
- name: industrial
dtype: string
- name: isolated_dwelling
dtype: string
- name: junction
dtype: string
- name: landuse
dtype: string
- name: leisure
dtype: string
- name: locality
dtype: string
- name: man_made
dtype: string
- name: military
dtype: string
- name: municipality
dtype: string
- name: natural
dtype: string
- name: neighbourhood
dtype: string
- name: office
dtype: string
- name: place
dtype: string
- name: postcode
dtype: string
- name: province
dtype: string
- name: quarter
dtype: string
- name: railway
dtype: string
- name: region
dtype: string
- name: residential
dtype: string
- name: retail
dtype: string
- name: road
dtype: string
- name: shop
dtype: string
- name: square
dtype: string
- name: state
dtype: string
- name: state_district
dtype: string
- name: suburb
dtype: string
- name: tourism
dtype: string
- name: town
dtype: string
- name: village
dtype: string
- name: boundingbox
sequence: string
- name: camera
dtype: string
- name: code
dtype: string
- name: date
dtype: string
- name: dimensions
sequence: int64
- name: display_name
dtype: string
- name: domain
dtype: string
- name: lat
dtype: string
- name: latlong
sequence: float64
- name: licence
dtype: string
- name: lon
dtype: string
- name: origin
sequence: float64
- name: osm_id
dtype: int64
- name: osm_type
dtype: string
- name: patch_centroid_x
dtype: float64
- name: patch_centroid_y
dtype: float64
- name: patch_centroid_z
dtype: float64
- name: place_id
dtype: int64
- name: time
dtype: string
- name: unit_system
dtype: string
- name: zone
dtype: string
- name: class_rep
struct:
- name: class_repartition
struct:
- name: agricultural land
dtype: float64
- name: bare soil
dtype: float64
- name: brushwood
dtype: float64
- name: building
dtype: float64
- name: clear cut
dtype: float64
- name: coniferous
dtype: float64
- name: deciduous
dtype: float64
- name: greenhouse
dtype: float64
- name: herbaceous vegetation
dtype: float64
- name: impervious surface
dtype: float64
- name: ligneous
dtype: float64
- name: mixed
dtype: float64
- name: other
dtype: float64
- name: pervious surface
dtype: float64
- name: plowed land
dtype: float64
- name: snow
dtype: float64
- name: swimming pool
dtype: float64
- name: vineyard
dtype: float64
- name: water
dtype: float64
- name: prompt
dtype: string
- name: clip_label
dtype: string
- name: clip_embeddings
sequence: float32
splits:
- name: train
num_bytes: 26969627407.0
num_examples: 61712
download_size: 26975935912
dataset_size: 26969627407.0
tags:
- IGN
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
---
# Dataset Card for "FLAIR_OSM_CLIP"

Dataset for the Seg2Sat model: https://github.com/RubenGres/Seg2Sat

Derived from [FLAIR#1](https://huggingface.co/datasets/IGNF/FLAIR) train split.

This dataset incudes the following features:
- **image**: FLAIR#1 .tif files RBG bands converted into a more managable jpg format
- **segmentation**: FLAIR#1 segmentation converted to JPG using the [LUT from the documentation](https://ignf.github.io/FLAIR/index_fr.html)
- **metadata**: OSM metadata for the centroid of the image
- **clip_label**: [CLIP ViT-H](https://github.com/openai/CLIP) description
- **class_rep**: ratio of appearance of each class in the segmentation
- **prompt**: Prompt used for the Seg2Sat model
- **clip_embeddings**: Embedding using openai/clip-vit-base-patch32 model

<!---
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
--->

原始信息汇总

数据集概述

数据集名称

• FLAIR_OSM_CLIP

数据集大小

• 10K<n<100K

数据集特征

1. image
   • 类型: 图像
2. segmentation
   • 类型: 图像
3. metadata
   • 结构:
     • address
       • 子结构:
         • ISO3166-2-lvl4: 字符串
         • ISO3166-2-lvl6: 字符串
         • ISO3166-2-lvl7: 字符串
         • aerialway: 字符串
         • aeroway: 字符串
         • allotments: 字符串
         • amenity: 字符串
         • bridge: 字符串
         • building: 字符串
         • city: 字符串
         • city_district: 字符串
         • club: 字符串
         • commercial: 字符串
         • country: 字符串
         • country_code: 字符串
         • county: 字符串
         • craft: 字符串
         • emergency: 字符串
         • farm: 字符串
         • farmyard: 字符串
         • hamlet: 字符串
         • healthcare: 字符串
         • highway: 字符串
         • historic: 字符串
         • house_number: 字符串
         • industrial: 字符串
         • isolated_dwelling: 字符串
         • junction: 字符串
         • landuse: 字符串
         • leisure: 字符串
         • locality: 字符串
         • man_made: 字符串
         • military: 字符串
         • municipality: 字符串
         • natural: 字符串
         • neighbourhood: 字符串
         • office: 字符串
         • place: 字符串
         • postcode: 字符串
         • province: 字符串
         • quarter: 字符串
         • railway: 字符串
         • region: 字符串
         • residential: 字符串
         • retail: 字符串
         • road: 字符串
         • shop: 字符串
         • square: 字符串
         • state: 字符串
         • state_district: 字符串
         • suburb: 字符串
         • tourism: 字符串
         • town: 字符串
         • village: 字符串
     • boundingbox: 字符串序列
     • camera: 字符串
     • code: 字符串
     • date: 字符串
     • dimensions: 整数序列
     • display_name: 字符串
     • domain: 字符串
     • lat: 字符串
     • latlong: 浮点数序列
     • licence: 字符串
     • lon: 字符串
     • origin: 浮点数序列
     • osm_id: 整数
     • osm_type: 字符串
     • patch_centroid_x: 浮点数
     • patch_centroid_y: 浮点数
     • patch_centroid_z: 浮点数
     • place_id: 整数
     • time: 字符串
     • unit_system: 字符串
     • zone: 字符串
4. class_rep
   • 结构:
     • class_repartition
       • 子结构:
         • agricultural land: 浮点数
         • bare soil: 浮点数
         • brushwood: 浮点数
         • building: 浮点数
         • clear cut: 浮点数
         • coniferous: 浮点数
         • deciduous: 浮点数
         • greenhouse: 浮点数
         • herbaceous vegetation: 浮点数
         • impervious surface: 浮点数
         • ligneous: 浮点数
         • mixed: 浮点数
         • other: 浮点数
         • pervious surface: 浮点数
         • plowed land: 浮点数
         • snow: 浮点数
         • swimming pool: 浮点数
         • vineyard: 浮点数
         • water: 浮点数
5. prompt
   • 类型: 字符串
6. clip_label
   • 类型: 字符串

数据集拆分

• train
  • 数据量: 61712个样本
  • 数据大小: 26842989610.0字节

下载大小

• 26786210818字节

数据集大小

• 26842989610.0字节

标签

• IGN

配置

• default
  • 数据文件:
    • 拆分: train
    • 路径: data/train-*