open-solution-mapping-challenge

open-solution-mapping-challenge copied to clipboard

jakubczakon

MulticlassLabeler returns different shapes

1

MulticlassLabeler usually (almost always) returns: `(2,300,300)` but sometimes you get `(1,300,300)` which of course causes trouble. It happens because of the `mask.max()` in the following implementation. ``` def label_multiclass_image(mask): labeled_channels...

jakubczakon

Funky augmentation where small buildings are added randomly as long as they are not touching any real image

jakubczakon

jakubczakon

jakubczakon

We may be training on too large images. Go through: - https://medium.com/mlreview/a-guide-to-receptive-field-arithmetic-for-convolutional-neural-networks-e0f514068807 - https://arxiv.org/abs/1701.04128 Decide whether we are training on correct size images: - Mayby smaller crops are enough in...

jakubczakon

kamil-kaczmarek

Models are performing badly on small images

jakubczakon

Class instances that are close are connected

jakubczakon

One of the options to speed-up data loaders is to: - transform your images and target masks into one large HDF5 file - refactor/add pytorch Datasets that read from the...

jakubczakon