Dear,

Your help would be appreciated. I wanna run your published code, but I'm facing such this error. How can I solve the error? I'm trying to run your first suggestion to execute your code on terminal. I'm running this below command as you said in README.md:

python test_batch.py --trainer UNIT --config ./configs/unit_img2dsm_folder.yaml --a2b 1 --input ./dataset/testA/ --output_folder ./results/outputs --checkpoint ./outputs/unit_img2dsm_folder/checkpoints/gen_00700000.pt

but unfortunately, I'm facing the mentioned error:
error: unrecognized arguments: --trainer UNIT

Dear,

How can I predict an RGB input image through this trained model "UIMG2DSM"? is there any python script for prediction purpose? Is it possible to use the trained model to predict a DSM from the RGB input image? how much pixel size would be suitable for the RGB input image? Your response would be appreciated.

Thanks,
Abbas Salehi

Thank you for the well-organized and reproducible code. I was able to run your model for inferencing on my own high-resolution satellite image data and got results. I am curious however if the elevation values are actually meaningful? What are the units of measure? Sorry if this information was already given somewhere and I missed it, but I would like to determine if the output is usable.

Dear,

I re-trained the network by adding my own data ( 217 optical and DSMs images) to your data, and then I got this below DSMs you can see the optical data and resulted DSMs. I do not know why the results (DSMs) are not accurate enough?! what is the reason that network generates such this bad DSMs? your suggestions would be appreciated.

optical images

DSM images