this project is based on labeme ,PyQT5 and pygdal. so the python3.6 env must be download and installed in the current folder.
you can using annaconda to download the python3.6 env and install PyQT5, gdal youself, and put it in the current folder
we provide a package
python3.7 + PyGdal + PyQT5,the extraction code is 3crv
After you download it and unzip to your root folder. you can run the RSLabel.exe
or you can download the binary from here rslabel-install the extraction code is lpps
-
build overviews for images.
for big images(larger than 300M), building overviews is highly recommended.
you can chose build overviews for single file or a folder
-
open the selected folder or a file
-
begin to draw or edit the polygon using toolbar buttons
-
when drawing, you can use right click to revoke the last point
-
when paning, the draw and edit action are disabled.
-
when drawing or editing, you can press CTRL button and drag the mouse to pan
- Support 32bit and 16bit RS IMAGE.
display single 16bit image with scalar remapping
display single 16bit image with pseudo-colour
- Fully compatible with LabelMe operating methods and provide other utilities to do with remote sensing image.
- Support Very large RS IMAGE
- Supporting infinite zooming
this tool is very convinient to label object with an angle. look that
you can export the result as VOC or COCO dataset's format , and use tensorflow or other deep learning framework to train the data.
When marking an image, if the image contains geographic coordinates, you can use it as a layer on the cesium digital earth, but this feature may affect the speed of the software, you can also remove this plug-in by removing the python/plugins directory under the root
,this tool allow you to add image to the cesium earth by add a tms imagery provider to cesium engine. but when the image is large , it will take long time to make the TMS tileset.
weixin 714601476