Unet with RGB, sampled points, hitmaps as imput channels (5 channels)

Run training

• Add the dataset (including inputs, outputs and Positions).
• Run generate_samples.py to generate samples under different sparse sampling rate.
• Run los.py to generate losmap.
• Run hit_batch.py to generate hitmap.
• Modify dataset.py and train.py to accomadate your feature selection.
• Run train.py to start training.

Note: Remember to update input channel number and feature dir when you need :)