How do I read multiple input images and one label at once when reading a dataset from disk?

General Discussion
#1

I have a dataset, its directory structure like this：
dataset/
1/
ldr1.jpg
ldr2.jpg
ldr3.jpg
hdr1.hdr
…
…
73/
ldr1.jpg
ldr2.jpg
ldr3.jpg
hdr1.hdr
As you see, each directory has three ldr image as input and one hdr image as label (groud truth). This is not like a traditional image classification problem, where one input corresponds to one label. In here, three input corresponds to one label.
What’s more, I have need to use some data augment to this three input ldr images. What should I do to use tensorflow to deal with this dataset?

1 Like
#3