Loss of information: upscaling after downscaling in get_item

[remtav]

deeplabv3/datasets.py

Line 74 in 415d983

scale = np.random.uniform(low=0.7, high=2.0)

In get_item, you resize image to 50% of original dimensions. Then, you randomly scale downscaled image to 70-200%. In the case of upsamplig to 200%, information about original image has been lost. Wouldn't it be best to work with original image if upscaling by factor of <100% ?