dcgan

The paper proposes a comprehensive data generation strategy combining traditional augmentation, CutMix, MixUp, and DCGAN-generated images to address data insufficiency and class imbalance for olive tree disease classification, evaluated using EfficientNetV2B0 and showing notable accuracy improvements.

Demonstrates running a DCGAN with 12.6M int8 quantized parameters on a low-cost RISC-V microcontroller (CH32H417), generating 64x64 cat faces in 26 seconds using pure C inference and quantum entropy sampling.

Trained a DCGAN from scratch on 350 photos of a red solo cup taken with an iPod touch 4, producing results reminiscent of early DALL-E.