Feature/fes/region filtering #153
Conversation
| "use_rds,geo_extension", | ||
| ([True, None], [False, ".gpkg"], [False, ".geojson"], [False, ".shp"]), | ||
| ) | ||
| def test_empty(self, tmp_path, use_rds, geo_extension): |
There was a problem hiding this comment.
I ran into an error with an empty GDF, this test replicated the error and then started passing when the error was fixed.
| Returns: | ||
| gpd.GeoDataFrame: Detections in the requested CRS | ||
| RegionDetections: Detections in the requested CRS |
There was a problem hiding this comment.
Just updating the docstring to match what the code is doing
| mask_iterator: Iterable[np.ndarray], | ||
| threshold: float = 0.4, | ||
| ) -> List[RegionDetectionsSet]: | ||
| ) -> Union[List[RegionDetectionsSet], List[gpd.GeoDataFrame]]: |
There was a problem hiding this comment.
Not a change in behavior, just correcting the hint to match the behavior
| (x0, y0) center of the ellipse in x and y (pixels) | ||
| axes (Tuple[int, int]): | ||
| (a, b) semi-major and semi-minor axis lengths in pixels | ||
| (a, b) x and y axis lengths (before rotation) in pixels |
There was a problem hiding this comment.
A more correct description
| detection_overlay = Image.new("RGBA", image.size, (0, 0, 0, 0)) | ||
| detection_draw = ImageDraw.Draw(detection_overlay, "RGBA") | ||
| for idx, row in gdf1.iterrows(): | ||
| has_match = row["unique_ID"] in gdf2["unique_ID"].values |
There was a problem hiding this comment.
The "unique_id" column gets created whenever merge() is done on a RegionDetectionsSet. So if the detections are being read from a .gpkg/.geojson file is it expected by the file to have the "unique_ID" column? If so should that be specified in the documentation?
There was a problem hiding this comment.
I'm not sure, the files that I was using (documented in the PR description) came from TDF and had that column so I assumed it would generally be available. I suppose
- Is there a better column available?
- If not, we can add
id_columnas an argument withunique_IDas the default
There was a problem hiding this comment.
Ah okay. Any files created using TDF is guaranteed to have the unique_ID column. I was only asking for use cases where this could get input vector files generated outside of TDF. But I guess there isn't a possibility of that happening.
If needed, I think it would be good to add id_column as an argument with unique_ID default like you suggested, just to keep it more flexible.
3 participants
Creates a visualization that allows you to see which detections were filtered out by a mask and why (by checking which detections different components of the mask overlaps with.
Tested using this code which resulted in images like this.