• Aggadi Gnanesh

Jewelry image generation from real-world celebrity images often suffers from global consistency issues using existing GAN-based methods. This paper proposes a conditional coordinate GAN (COCO-GAN) model that generates high-resolution (256×256 and above) lucid jewelry images through spatially coordinated patch-based generation. Leveraging rough-to-detailed multi-stage training, the model synthesizes images with improved coherence and detail. Evaluations on the Celeb Jewelry Dataset validate COCO-GAN’s superior image generation capability, including its unique ability to perform beyond-boundary and patch-guided generation, all while reducing memory consumption significantly.

GAN, COCOGAN, Image Generation, Patch-based Training, Jewelry Synthesis, Coordinate-aware GAN, Deep Learning

"Cloudy-to-Lucid Jewelry Image Generation with COCOGAN", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.12, Issue 7, page no.a913-a920, July-2025, Available :http://www.jetir.org/papers/JETIR2507093.pdf

"Cloudy-to-Lucid Jewelry Image Generation with COCOGAN", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.12, Issue 7, page no. ppa913-a920, July-2025, Available at : http://www.jetir.org/papers/JETIR2507093.pdf