I tried to retrain a model from this github but I encounter a problem of creating hdf5 for it. So far i only know that hdf5 should contain images path and possible label/class of it. But on my case, it has a combination of images and matrix of light estimation. Is this possible to achieve or did I make wrong interpretation of the instruction. Please help me understand this
This is one of the light estimation matrix that they provide
0.300794 0.121209 -0.115402 0.073493 -0.126265 -0.035606 0.013735 -0.018538 -0.055469 0.202293 0.105253 -0.058834 0.047335 -0.097811 -0.032969 0.010878 -0.024713 -0.028035 0.150027 0.096441 -0.049527 0.032507 -0.088273 -0.033228 0.007316 -0.015788 -0.021543