Object detection using multibands

[greenhawktobias]

I will try object detection on trees with my own dataset preferable using 4 bands (RGB+elevation)
Looking at the Potsdam example "cowc_potsdam.py" it seems like the indata is 4 channels (RBG+IR) but the object detection is done with only 3 bands.

raster_source = RasterioSourceConfig(
            uris=[raster_uri], channel_order=[0, 1, 2])

Is it possible to do object detection using 4 bands? I am using raster vision version Release 0.13.0 with Docker.

[greenhawktobias]

This seems to be an updated version of the Potsdam code:
https://github.com/azavea/raster-vision/blob/master/rastervision_pytorch_backend/rastervision/pytorch_backend/examples/object_detection/cowc_potsdam.py#L86
Is it best to use this code with the latest tag of RV?

[greenhawktobias]

Traceback (most recent call last):
  File "/opt/conda/lib/python3.9/runpy.py", line 197, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "/opt/conda/lib/python3.9/runpy.py", line 87, in _run_code
    exec(code, run_globals)
  File "/opt/src/rastervision_pipeline/rastervision/pipeline/cli.py", line 250, in <module>
    _main()
  File "/opt/src/rastervision_pipeline/rastervision/pipeline/cli.py", line 246, in _main
    main()
  File "/opt/conda/lib/python3.9/site-packages/click/core.py", line 1130, in __call__
    return self.main(*args, **kwargs)
  File "/opt/conda/lib/python3.9/site-packages/click/core.py", line 1055, in main
    rv = self.invoke(ctx)
  File "/opt/conda/lib/python3.9/site-packages/click/core.py", line 1657, in invoke
    return _process_result(sub_ctx.command.invoke(sub_ctx))
  File "/opt/conda/lib/python3.9/site-packages/click/core.py", line 1404, in invoke
    return ctx.invoke(self.callback, **ctx.params)
  File "/opt/conda/lib/python3.9/site-packages/click/core.py", line 760, in invoke
    return __callback(*args, **kwargs)
  File "/opt/src/rastervision_pipeline/rastervision/pipeline/cli.py", line 170, in run
    cfgs = get_configs(cfg_module, runner, args)
  File "/opt/src/rastervision_pipeline/rastervision/pipeline/cli.py", line 74, in get_configs
    cfgs = _get_configs(runner, **args)
  File "code/obj.py", line 59, in get_config
    dataset = DatasetConfig(
  File "pydantic/main.py", line 342, in pydantic.main.BaseModel.__init__
pydantic.error_wrappers.ValidationError: 1 validation error for DatasetConfig
class_config
  value is not a valid dict (type=type_error.dict)

[AdeelH]

The error is about class_config. You are passing a class_config to DatasetConfig without having defined a class_config variable. The reason this runs at all is the from rastervision.core.data import * line which is importing the class_config module.

So you just need to define a class_config variable.

[greenhawktobias]

yes, I corrected that and it seems to work now. Thanks!

I am testing RasterVision for urban tree detection in a municipality together with a group of GIS-engineers. Maybe something worth to share on "show and tell" under "discussions"? Rastervision is a very good framework, would be nice to share some results.

[AdeelH]

yes, I corrected that and it seems to work now. Thanks!

Awesome!

I am testing RasterVision for urban tree detection in a municipality together with a group of GIS-engineers. Maybe something worth to share on "show and tell" under "discussions"? Rastervision is a very good framework, would be nice to share some results.

That would be most welcome! Please do share!

[AdeelH]

I am testing RasterVision for urban tree detection in a municipality together with a group of GIS-engineers. Maybe something worth to share on "show and tell" under "discussions"? Rastervision is a very good framework, would be nice to share some results.

That would be most welcome! Please do share!

@greenhawktobias, it might also be possible to turn this into a joint blog or case study of sorts and publish it online. Is that something you would be interested in? If you can share your email (or email me at ---) we can explore this possibility.