As the title says, I tried to grow apples using DCGAN. Click here for the DCGAN paper (https://arxiv.org/abs/1511.06434) I don't know if it's dirty and helpful, but I also tried implementing it → Implementation as a result
Has been generated. It's almost an apple.
The structure of the Generator is as follows.
Generates $ Z $ (leftmost) that follows a uniform distribution, repeats convolution, and generates an image.
It's dirty, but I implemented it in here. By the way, I referred to here.
For the data, I used 498 apple images from here in Kaggle's dataset. I turned 1000 epochs with a batch size of 128. You can generate it nicely.
Recommended Posts