Several issues on 3_model.ipynb

[kate-berezn]

I can run this notebook to the end as I faced with several issues:

1. Variables (x0, y0,rx and ry) in the cell with content below are not defined

filter pipes to ones inside the image

import shapely
box = shapely.geometry.box(
minx=x0,
miny=y0,
maxx=x0+rxdata.shape[1],
maxy=y0+rydata.shape[2],
)

2. variables X_val2 and y_val2 are not defined so the code below fails.
   I tried to guess which data I should pass and just passed "X_val, y_val" for now but it does not look right (as results of training/test is strange)

pretrain/test

history = model.fit_generator(
datagen.flow(X_train, y_train, batch_size=batch_size),
samples_per_epoch=len(X_train)*4,
verbose=0,
nb_epoch=60,
validation_data=[X_val2,y_val2],

Any idea on how to work-around or fix this?

[wassname]

Copied from email

Sorry I'm late in replying, here and on github. I'm flat out at work, and haven't looked at this project for years.

I'm also afraid it's a bit messy... like not defining x0. Sorry about that, I just wanted to share it, even if I couldn't find the time to clean it up well.

It looks like I was running the cells out of order, and getting x0, from the tfw or tiff world file. But it looks like it's just:
x0 and y0 are Easting and Northing of the upper left pixel (0,0 in image coordinates). rx and ry are x-pixel resolution and y-pixel resolution.