Machine Learning

1 readers

1 users here now

Community Rules:

Be nice. No offensive behavior, insults or attacks: we encourage a diverse community in which members feel safe and have a voice.
Make your post clear and comprehensive: posts that lack insight or effort will be removed. (ex: questions which are easily googled)
Beginner or career related questions go elsewhere. This community is focused in discussion of research and new projects that advance the state-of-the-art.
Limit self-promotion. Comments and posts should be first and foremost about topics of interest to ML observers and practitioners. Limited self-promotion is tolerated, but the sub is not here as merely a source for free advertisement. Such posts will be removed at the discretion of the mods.

founded 1 year ago

MODERATORS

communick@academy.garden

[D] About data augmentation (alien.top)

submitted 1 year ago by Cautious_Employer298@alien.top to c/machinelearning@academy.garden

2 comments fedilink hide all child comments

I have an assignment where we have to use DCGAN to generate images of a bus. But the dataset they have given, on purpose has very less images (108) and the images also have very low light (taken at night). I tried to train using a very simple architecture, but even after 1000 epochs, the images are not meaningful. What can I do to increase the lighting, and what type of data augmentation technique should I use?

The dataset given to me

top 2 comments

sorted by: hot top controversial new old

[–] Infamous-Bank-7739@alien.top 1 points 1 year ago (1 children)

Note that 108 is a very small set. Considering transfering a pre-learned net or other means.

But to answer your question: I've had good success with DiffAug. It's quite straightforward to implement as well. ADA is another popular one that is bit more complex.

[–] Cautious_Employer298@alien.top 1 points 1 year ago

interesting, thanks