geo-deep-learning
geo-deep-learning copied to clipboard
NRCan
Deep learning applied to georeferenced datasets
A patch stride is present in our Tiler class attribute, but it not yet implemented: https://github.com/NRCan/geo-deep-learning/blob/develop/tiling_segmentation.py#L71 To complete the implementation, it should be used to set the stride for the...
AOI tests dont go as wanted, some of them crash inside the AOI class initialization without raising the good error.  * `test_multiband_input_band_selection_from_letters` * `test_multiband_input_band_selection_too_many` * the line 120-121 are...
Although geo-deep-learning is meant to deal only with ground truth files containing polygons or multipolygons, no warning is issued if ground truth data contains such vector data. The following ground...
Current gpkg inference of the GDL can contain such objects like GeometryCollection, LineString, etc. We need to add some extra geometry checks in the post-processing pipeline to get only Polygon...
 `save_weights_dir`, `state_dict_path`, `checkpoint_dir` all mean roughly the same, namely, the location/path of where the trained model is stored. Can we reduce the different names to just one? Three different...
If we set custom local dir as `mlflow.yaml`'s `uri`, it will raise an exception.  The culprit is `mlflow.set_tracking_uri()` function.  We can fix it by identifying `uri` being a...
The images that are passed to the models do not always have a uniform, quadrilateral shape. This causes that sometimes, patches completely void of data are passed to the model,...
Although undesirable, there may be rare cases where the DigitalGlobe archive has already been enhanced before it enters the inference process. In fact, there is a optional parameter when ordering...
After changing the `get_key_def` function, I notice some `get_key_def` function with the parameter `to_path=True` have also the parameter `expected_type=str` but with the new modification those one need to be change...
[Validation script in production ](https://github.com/remtav/geo-deep-learning/blob/222-stac-item-input/validate_segmentation.py) is failing ```python [[36m2023-01-18 14:32:59,997[0m][[34mroot[0m][[32mINFO[0m] - Overwritten parameters in the config: inference.checkpoint_dir=[...]/operationalization/checkpoints/,inference.input_stac_item=https://datacube-stage.services.geo.ca/api/collections/worldview-2-ortho-pansharp/items/ON17-052934190130_01_P001-WV02,inference.output_name=ON17-052934190130_01_P001-WV02_ROAI,inference.state_dict_path=https://datacube-[...]/models/pl_VA_95imgs_RGBN_ROAI_20220925.pth.tar,mode=validate,postprocess.output_name=ON17-052934190130_01_P001-WV02,postprocess.reg_cont.cont_image=[...]/singularity_images/gdl-cuda11_v2.3.3-prod.sif,postprocess.root_dir=[...]operationalization/inferences/[0m [[36m2023-01-18 14:32:59,998[0m][[34mroot[0m][[32mINFO[0m] - ----------------------------------------- Let's start validate for segmentation !!! -----------------------------------------[0m [[36m2023-01-18 14:33:00,812[0m][[34mroot[0m][[32mINFO[0m]...