Opening open-source SpaceNet 6 expanded dataset

Question

So I am trying to access the open-source SAR dataset SpaceNet 6: Expanded Dataset. I particularly need the SAR-SLC data as I want to have the raw complex-valued dataset.

I downloaded the dataset successfully using aws. Now the issue arises when I use python to open the .tif files. I tried OpenCV, matplotlib, PIL and skimage. Here the errors I get:

• OpenCV: [ WARN:[email protected]] global /io/opencv/modules/imgcodecs/src/grfmt_tiff.cpp (462) readData OpenCV TIFF: TIFFRGBAImageOK: Sorry, can not handle images with 32-bit samples.
• Matplotlib: cannot identify image file '<path to .tif file>'.
• PIL: cannot identify image file '<path to .tif file>'.
• skimage: TiffPage 0: data type not supported: 532.

Has someone already succeeded in opening this library?

---

I've seen that this code uses a library called solaris. I tried installing it but it requires TensorFlow version to be tf v1.13 and I already have version 2 and I need it. I tried creating a new environment to install it from scratch. As the documentation says I need to install GDAL so at this point why not? sudo apt-get install libgdal-dev but then I still got errors like:

ERROR: Cannot install solaris==0.0.1, solaris==0.0.2, solaris==0.0.3, solaris==0.1.0, solaris==0.1.1, solaris==0.1.2, solaris==0.1.3, solaris==0.2.0, solaris==0.2.1, solaris==0.2.2, solaris==0.3.0 and solaris==0.4.0 because these package versions have conflicting dependencies.

The conflict is caused by:
    solaris 0.4.0 depends on torch>=1.3.1
    solaris 0.3.0 depends on torch>=1.3.1
    solaris 0.2.2 depends on torch>=1.3.1
    solaris 0.2.1 depends on torchvision>=0.4.2
    solaris 0.2.0 depends on torchvision>=0.3.0
    solaris 0.1.3 depends on torchvision>=0.3.0
    solaris 0.1.2 depends on torchvision>=0.3.0
    solaris 0.1.1 depends on torchvision>=0.3.0
    solaris 0.1.0 depends on torchvision>=0.3.0
    solaris 0.0.3 depends on torchvision>=0.3.0
    solaris 0.0.2 depends on torchvision>=0.3.0
    solaris 0.0.1 depends on torch>=1.0.1

And it was a NEW environment without tf nor pytorch installed.

This, although a crapy solution (Do I really need to create a new env, install tf and pytorch and lot of dependencies, install GDAL, etc to open an image?) at this point I am Ok with whatever works.