Marcus D. Bloice

Hi there @NithyaMogane-TomTom Actually, this is not supported. The skew operations maintain the image size, as different skew magnitudes/angles would result in many different image sizes due to the stochastic...

Hi @AKJDM, are you perhaps working with PNGs that have an alpha channel? You could try setting the save file format to PNG so that they are not converted to...

Hi @yesufeng, thanks for pointing this out, however I cannot seem to recreate this when I test it. PIL doesn't seem to drop channels by default, and if I explicitly...

Hi @marcojulioarg, yeah you need to pass and return a list now and not a PIL image. The documentation may be slightly out of date regarding this actually. The change...

Hi @KokeCacao, ok thanks for making the issue and finding this bug. I will take a look at that as soon as I can.

Hi @ajuric, thanks for opening the issue - no currently you cannot do either of those features, however I like the idea especially regarding the semantic segmentation. I will definitely...

@ajuric thanks to a PR by @ajuric this is now possible for rotations, Augmentor version 0.2.4 now includes this feature. As for the different colours for the ground truth and...

Hi @ssundar6087, sure let's try to fix the problem. This has something to do with how Augmentor and PIL handle saving images of different formats. Do your original files have...

Also, are you maybe initialising the pipeline using the `save_format` parameter like so: ```python p = Augmentor.Pipeline("/tmp/images/", save_format="BMP") ``` or something similar? If you force a format using `save_format`, then...

Ok, it is rather strange as this works on my end. Another question: are the ground truth images also in the same format?