Broadcasting of remote sensing resources (2) | 11 classic remote sensing data sets

Time:2021-4-8

Abstract: semantic segmentation is the most commonly used method for land classification based on remote sensing images. Following the previous training course of land classification model, several major open remote sensing data sets are organized.

Original: hyperai super nerve

Key words: remote sensing data set ﹣ semantic segmentation ﹣ machine vision


In the last issueRemote sensing resources broadcast (Part 1): training land classification model with open source codeIn this paper, we introduce in detail the common methods of land classification using remote sensing images, and how to use deeplab-v3 + for the specific training process and code of land classification.

Tutorial address:

https://openbayes.com/console…

Broadcasting of remote sensing resources (2) | 11 classic remote sensing data setsExamples of original remote sensing image (Part 1) and 7 classification image (Part 2)

In addition to classification and recognition, the common scenes of remote sensing images include:

  • Super resolution reconstruction
  • Fast processing of multi source remote sensing images
  • Distributed storage of remote sensing image

In this issue, we have sorted out 11 remote sensing data sets. There are at least two types of detected targets and at most 45 types. We can provide more sufficient “ammunition” for your model training. Please use it as needed.

Part 1: for 2-5 classification problems

Ucas-aod remote sensing image data set

UCAS AOD remote sensing image data set is used for aircraft and vehicle detection.

Specifically, the aircraft data set includes 600 images and 3210 aircraft, while the vehicle data set includes 310 images and 2819 vehicles. All the images are carefully selected to make the objects in the data set evenly distributed.

Broadcasting of remote sensing resources (2) | 11 classic remote sensing data setsExamples of vehicle (a) and aircraft (b) target detection

The data set was first published by the University of the Chinese Academy of Sciences in 2014 and supplemented in 2015. The related papers include orientation robust object detection in aerial images using deep convolutional neural network.

Here are the details of the dataset:

Ucas-aod remote sensing image data set

Issued by:University of Chinese Academy of Sciences

Update time:Issued in 2014, supplemented in 2015

Included quantity:600 aircraft & 310 vehicle images

Image source:Google Earth Satellite Images

Data format:.png

Picture size:1280*659

Data size:3.48GB

Number of categories:Category 2

Download address:_https://hyper.ai/datasets/5419_

Inria Aerial Image Labeling Dataset

INRIA aerial image labeling dataset is a remote sensing image dataset for urban building detection,Its tags are divided into building and not building, which are mainly used for semantic segmentation.

Broadcasting of remote sensing resources (2) | 11 classic remote sensing data sets

Here are the details of the dataset:

INRIA aerial image labeling dataset

Issued by:INRIA (National Institute of information and automation, France)

Release time: 2017

Included quantity:360 images

Data format:GeoTiff

Image size:5000*5000

Data size:69GB

Number of categories:Class 2

Release time:2017

Download address:_https://hyper.ai/datasets/5428_

RSOD dataset object detection dataset

RSOD dataset is used to detect objects in remote sensing images,It includes four types of targets: aircraft, playground, overpass and oil tank,The quantities are as follows:

  • Aircraft: 446 drawings, including 4993 aircraft;
  • Playground: 189 pictures, including 191 playgrounds;
  • Overpass: 176 drawings, including 180 overpasses;
  • Oil drums: 165 drawings, including 1586 oil drums.

Broadcasting of remote sensing resources (2) | 11 classic remote sensing data sets

Here are the details of the dataset:

RSOD Dataset

Issued by:WuHan University

Release time: 2015

Included quantity:976 images

Data format:.jpg

Data size:324.96MB

Number of categories:Category 4

Download address:_https://hyper.ai/datasets/5425_

Part 2: for 5-10 classification problems

Rsscn7 dataset remote sensing image dataset

Rsscn7 dataset contains 2800 remote sensing images from seven typical scene categoriesGrasslands, forests, farmland, car parks, residential areas, industrial areas and rivers and lakes,Each category contains 400 images, which are sampled based on four different scales.

The pixel size of each image in the dataset is 400 * 400. The diversity of scene images makes it more challenging. These images come from different seasons and weather changes and are sampled in different proportions.

Broadcasting of remote sensing resources (2) | 11 classic remote sensing data sets

Here are the details of the dataset:

RSSCN7 Dataset

Issued by:WuHan University

Release time: 2015

Included quantity:2800 images

Data format:.jpg

Image size:400*400

Data size:348.02MB

Number of categories:7 categories

Download address:_https://hyper.ai/datasets/5440_

NWPU vhr-10 remote sensing data set for geospatial object detection

NWPU vhr-10 dataset is a 10 level remote sensing data set for space object detection. It has 650 images containing targets and 150 background images, a total of 800. It is cut from Google Earth and vaihingen data sets, and then manually annotated by experts.

Its target categories include aircraft, ships, oil tanks, baseball fields, tennis courts, basketball courts, track and field fields, ports, bridges and cars, a total of 10 categories.

Broadcasting of remote sensing resources (2) | 11 classic remote sensing data sets

Here are the details of the dataset:

NWPU vhr-10 data set

Issued by:Northwest University of Technology

Release time:2014

Included quantity:800 images

Data format:.jpg

Data size:73MB

Number of categories:10 categories

Download address:_https://hyper.ai/datasets/5422_

Part 3: for 11-20 classification problems

Rsc11 dataset remote sensing image data set

Rsc11 dataset is a remote sensing image data set, which comes from the high-resolution remote sensing image of Google Earth,There are 11 kinds of scene images, including dense forest, sparse forest, grassland, port, high-rise building, low rise building, overpass, railway, residential area, road and storage tank.There are about 100 pieces in each category, a total of 1232 pieces, and the spatial resolution is 0.2m.

The data set was released by the Chinese Academy of Sciences in 2015, mainly by Zhao Lijun.

Here are the details of the dataset:

Rsc11? Data set

Issued by:Chinese Academy of Sciences

Release time:2015

Included quantity:1232 images

Data format:.tif

Picture size:512*512

Data size:20.12MB

Number of categories:11 categories

Download address:_https://hyper.ai/datasets/5443_

Siri-whu remote sensing image data set

The siri-whu dataset contains 12 categories of scenes, with a total of 2400 images, 200 of which are in each category. The pixel size of each image is 200 * 200, and the spatial resolution is 2m.

The dataset resources are from Google Earth, mainly covering urban areas in China. The scene image dataset is designed by rs-idea group of Wuhan University.

Broadcasting of remote sensing resources (2) | 11 classic remote sensing data sets

Here are the details of the dataset:

Siri-whu remote sensing image data set

Issued by:WuHan University

Release time:2016

Included quantity:2400 images

Data format:.tif

Picture size:200*200

Data size:162.08MB

Number of categories:12 categories

Download address:_https://hyper.ai/datasets/5437_

Whu-rs19 dataset remote sensing image data set

Whu-rs19 dataset is a remote sensing image dataset, which contains 19 categories of scene images, a total of 1005, of which each category has 50. It can be used for scene classification and retrieval.

The data set was published by Wuhan University in 2011, and the related papers are satellite image classification via two layer sparse coding with biased image representation.

Broadcasting of remote sensing resources (2) | 11 classic remote sensing data sets

Here are the details of the dataset:

Whu-rs19 data set

Issued by:WuHan University

Release time:2011

Included quantity:1005 images

Data format:.tif

Picture size:600*600

Data size:99.54MB

Number of categories:19 categories

Download address:_https://hyper.ai/datasets/5434_

Part 4: for 20 + classification problems

UC Merced Land-Use DataSet

UC Merced land use dataset is a 21 level remote sensing data set of land use images for research, which is extracted from USGS national map urban area image series and used in urban areas all over the country.

The pixel resolution of the public domain image in this dataset is 1 foot (0.3 m), and the image pixel size is 256 * 256,There are 2100 scene images in 21 categories, including 100 in each category.

The 21 categories are:Agriculture, airplanes, baseball fields, beaches, buildings, bushes, dense houses, forests, expressways, golf courses, ports, intersections, medium-sized houses, mobile family parks, overpasses, parking lots, rivers, runways, sparse houses, oil tanks.

Broadcasting of remote sensing resources (2) | 11 classic remote sensing data sets

Here are the details of the dataset:

UC merged land use dataset

Issued by:UC Merced Vision&Learning Lab

Included quantity:2100 sheets

Data format:.png

Picture size:256*256

Data size:317.07MB

Number of categories:21 categories

Release time:2010

Download address:_https://hyper.ai/datasets/5431_

Aid dataset remote sensing image data set

Aid dataset is a remote sensing image data set, which contains 30 categories of scene images. Each category has about 220 – 420 images, with a total of 10000 images. The pixel size of each image is about 600 * 600.

The data set was published by Huazhong University of science and technology and Wuhan University in 2016, and the related paper is aid: a benchmark dataset for performance evaluation of aerial scene classification.

Broadcasting of remote sensing resources (2) | 11 classic remote sensing data sets

Here are the details of the dataset:

Aid remote sensing image data set

Issued by:Huazhong University of science and technology and Wuhan University

Included quantity:10000

Data format:.jpg

Picture size:600*600

Data size: 2.47GB

Number of categories:30 categories

Release time:2016

Download address:_https://hyper.ai/datasets/5446_

NWPU dataset remote sensing image data set

NWPU dataset contains 31500 images with 256 * 256 pixels, covering 45 scene categories, including 700 images in each category.

These 45 The three scene categories include airplane, airport, baseball field, basketball court, beach, bridge, jungle, church, circular farmland, cloud, business district, dense residence, desert, forest, expressway, golf course, ground track and field, port, industrial area, intersection, island, lake, grassland, medium-sized residence, mobile house park, mountain, overpass, palace, parking lot and railway , railway station, rectangular farmland, river, circular transportation hub, runway, sea, ship, snow mountain, sparse residence, stadium, water storage tank, tennis court, terrace, thermal power station and wetland.

Broadcasting of remote sensing resources (2) | 11 classic remote sensing data sets

The data set is published by Northwestern University of technology, and the related papers are remote sensing image scene classification: benchmark and state of the art.

Here are the details of the dataset:

NWPU dataset remote sensing image data set

Issued by:Northwest University of Technology

Included quantity:31500

Data format:.jpg

Picture size:256*256

Data size: 403.71MB

Number of categories:Class 45

Release time:2017

Download address:_https://hyper.ai/datasets/5449_

The above is the whole content of this resource package, you can also visitThe following links: https://hyper.ai/ Search for remote sensing data sets,You can get to all the resources.