Query on what data fields need to be appended in csv file

[MathewAaron]

I believe the csv file is appended by the following fields :

im_filename = "test.jpg"

latitude = -39.1

longitude = 71.2

grass_pixels = 1

grass_biomass = 2

clover_pixels = 3

clover_biomass = 4

brassica_pixels = 5

brassica_biomass = 6

weed_pixels = 7

weed_biomass = 8

total_vegetation_pixels = 9

total_biomass = 10

I need help in identifying where each parameter is computed from?

[skovsens]

The majority of these parameters are coming from the custom ros msg that holds the results for each data sample.

The image filename is generated somewhere else, and serves just as a way to debug. In case we see some weird results, it's good to be able to backtrack the results. Is it perhaps included in the custom ros msg per image sample already?

GPS-coordinates is already present in the custom msg.

Class-wise pixel counts are basically np.sum((semantic_segmentation_image == classID).astype('int32')) / im_size where classID corresponds to the numerical representations of the classes (fx soil=0, grass=1 etc..).
Perhaps this should be computed in the biomass predictor as well and put into the custom msg. If not, you can do it in this data saver by looping the classes through the code snippet above. I can see that I forgot to add "soil_pixels", but the same applies for that class.

Biomass metrics should have already been outputted by the biomass prediction node and appended to the custom msg.

Regarding the "total_x" metrics, that one is a sum of other metrics from above. "total_vegetation_pixels" is the sum of grass_pixels, clover_pixels, brassica_pixels and weed_pixels. Vise versa for "total_biomass"