Skip to content

whut2962575697/Awesome-Deep-Learning-of-Remote-Sensing

Repository files navigation

Awesome Deep Learning of Remote Sensing Awesome

In this project, we will open source some baseline codes for the remote sensing analysis task, such as semantic segmentation, scene classification, object detection, and image captioning, We will also collect some public datasets that can be used for remote sensing image research and analysis.

  • Public Remote Sensing Dataset
  • Baseline Codes (Semantic Segmentation/Scene Classification/Object Detection/Image Captioning)
  • OpenSoure Codes
  • Compitions About Remote Sensing

Public Remote Sensing Dataset

1.Semantic Segmentation

6 urban land cover classes, raster mask labels, 4-band RGB-IR aerial imagery (0.05m res.) & DSM, 38 image patches

categories
index label color
1 Impervious surfaces 255, 255, 255
2 Building 0, 0, 255
3 Low vegetation 0, 255, 255
4 Tree 0, 255, 0
5 Car 255, 255, 0
6 Clutter/background 255, 0, 0
Download

baiduyun password: 9enz

10 land cover categories from crops to vehicle small, 57 1x1km images, 3/16-band Worldview 3 imagery (0.3m-7.5m res.), Kaggle kernels

10 land cover classes, temporal stack of hyperspectral Sentinel-2 imagery (R,G,B,NIR,SWIR1,SWIR2; 10 m res.) for year 2017 with cloud masks, Official Slovenian land use land cover layer as ground truth.

180,748 corresponding image triplets containing Sentinel-1 (VV&VH), Sentinel-2 (all bands, cloud-free), and MODIS-derived land cover maps (IGBP, LCCS, 17 classes, 500m res.). All data upsampled to 10m res., georeferenced, covering all continents and meterological seasons, Paper: Schmitt et al. 2018

20 land cover categories by fusing three data sources: Multispectral LiDAR, Hyperspectral (1m), RGB imagery (0.05m res.)

16 land cover classes,4-band RGB-IR aerial imagery (4m res.) 8 patches of 7200x6800 for train and 2 patches of 7200x6800 for val and 10 patches of 7200x6800 for test

categories
index label color
1 水田 0,200,0
2 水浇田 150,250,0
3 旱耕地 150,200,150
4 园地 200,0,200
5 乔木林地 150,0,250
6 灌木林地 150,150,250
7 天然草地 250,200,0
8 人工草地 200,200,0
9 工业用地 200,0,0
10 城市住宅 250,0,150
11 村镇住宅 200,150,150
12 交通运输 250,150,150
13 河流 0,0,200
14 湖泊 0,150,200
15 坑塘 0,200,250
16 其他 0,0,0
Download

baiduyun password: o2fp

5 argriculture categories

5 land cover classes(greenland, building, waterbody, road and other), 5 rgb images(R,G,B; 1 m res.) for train and val, 3 rgb images for test

categories
index label gary
1 植被 1
2 建筑 2
3 水体 3
4 道路 4
5 其他 0
Download

baiduyun password: aaod

  • 初赛:10万高分光学影像和标注文件(一级分类(8类)),20万测试图片数据;
  • 复赛:10万高分光学影像和标注文件(二级分类(17类)),30万测试图片数据;
初赛 categories
index 一级标签 gary(百位数字)
1 水体 1
2 交通运输 2
3 建筑 3
4 耕地 4
5 草地 5
6 林地 6
7 裸土 7
8 其他 8
复赛 categories
index 二级标签 gary(十位及个位上的数字)
1 水体 01
2 道路 02
3 建筑物 03
4 机场 04
5 火车站 05
6 光伏 06
7 停车场 07
8 操场 08
9 普通耕地 09
10 农业大棚 10
11 自然草地 11
12 绿地绿化 12
13 自然林 13
14 人工林 14
15 自然裸土 15
16 人为裸土 16
17 其他 17
Download

baiduyun password: pdos

训练集包含140,000张分辨率为2m/pixel,尺寸为256256的JPG图片,一共7个类别,对应gt 0-6

categories
index label gary
1 建筑 0
2 耕地 1
3 林地 2
4 水体 3
5 道路 4
6 草地 5
7 其他 6
8 未标注区域 255
Download

baiduyun password: 7tcn

初赛训练集包含16017张分辨率为0.8m-2m/pixel,尺寸为256256的TIF图片,一共10个类别,对应gt 1-10
复赛训练集包含15904张分辨率为0.8m-2m/pixel,尺寸为256256的TIF图片,一共10个类别,对应gt 1-10

categories
index label gary
1 耕地 1
2 林地 2
3 草地 3
4 道路 4
5 城镇建设用地 5
6 农村建设用地 6
7 工业用地 7
8 构筑物 8
9 水域 9
10 裸地 10
Download

baiduyun password: td5k

2.Scene Classification

3.Object Detection

4.Image Captioning

Baseline Codes

Scene Classification

image_classification [pytorch]

Semantic Segmentation

image_seg [pytorch]
It is based on the codes of our Tianchi competition in 2021 (https://tianchi.aliyun.com/competition/entrance/531860/introduction).
In the competition, our team won the third place (please see Tianchi_README.md).

OpenSoure Codes

1.Semantic Segmentation

2.Scene Classification

3.Object Detection

4.Image Captioning

Compitions About Remote Sensing

2020

2019

2018

2017

About

Awesome Deep Learning of Remote Sensing

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages