reductionista commented on issue #14: Image loader python module, and demo notebook URL: https://github.com/apache/madlib-site/pull/14#issuecomment-504127456 > not sure if this is an error or not? > > ``` > import numpy as np > x1 = np.ones((3,4,5)) > y1 = np.ones((1,1)) > iloader = ImageLoader(num_workers=1, db_creds=db_creds) > iloader.load_np_array_to_table(x1, y1, 'test1', append=True, img_names=None) > > x1 > array([[[1., 1., 1., 1., 1.], > [1., 1., 1., 1., 1.], > [1., 1., 1., 1., 1.], > [1., 1., 1., 1., 1.]], > > [[1., 1., 1., 1., 1.], > [1., 1., 1., 1., 1.], > [1., 1., 1., 1., 1.], > [1., 1., 1., 1., 1.]], > > [[1., 1., 1., 1., 1.], > [1., 1., 1., 1., 1.], > [1., 1., 1., 1., 1.], > [1., 1., 1., 1., 1.]]]) > ``` > > ``` > madlib=# select * from test1; > id | x | y > ----+---------------------------------------------------+----- > 1 | {{1,1,1,1,1},{1,1,1,1,1},{1,1,1,1,1},{1,1,1,1,1}} | 1.0 > (1 rows) > ``` > > did not load the whole array Outer dimension of x must match the dimension of y; since there is only 1 label supplied here, so it matches that with the first image and ignores all the extra images. I can't recall if we are supposed to be supporting 1-hot encoded labels passed from the user. If we need to support that, then we can add that to our list of followup tasks after this merge; either way this specific case is a user error. For now, I will add a check for this, so we can give an error message instead of proceeding with invalid input.
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
