Hackathon2020-Pier24

Content for submission.

Index

Problem Workflow
Datasets
Tasks
References

Problem-workflow

Datasets

rio

Rio de Janeiro (Brazil) satellite imaging dataset for SpaceNet 1: Building Detection v1 challenge.

train:   3,8 band satellite images (raster) & building-footprints (vector) for Rio-de-Janeiro (2784 sq KM) collected by WorldView-2
  mosaic: 3 band mosaic of dataset & building-footprints
test:    3,8 band satellite images (raster) for Rio-de-Janeiro (2784 sq KM) collected by WorldView-2
labels:  Compressed 3,8 band 200m x 200m tiles with associated building foot print labels
landuse: Categorized land regions (shp/geojson) [22946 features].

landuse data has 18 regional classes.

$regex='"fclass": "(\w+)"'
$classes = Select-String -Path '.\landuse.geojson' -Pattern $regex -AllMatches | forEach {$_.matches.groups[0]} | forEach {$_.Value -replace $regex,'$1'} | Sort-Object | Get-Unique

allotments
cemetery
commercial
farm
forest
grass
heath
industrial
meadow
military
nature_reserve
orchard
park
quarry
recreation_ground
residential
retail
scrub