Enhancing Medical Image Generation with Advanced GAN Architectures
Budget: $10 – $20 USD
I need a freelancer to enhance my existing WGAN-GP implementation in PyTorch for OrganAMNIST image generation using MedMNIST. The current Google Colab code produces low-quality outputs, lacking diversity and suffering from training instability.
Key Requirements:
- Using multiple GAN architectures (WGAN-GP, DCGAN, StyleGAN, Conditional GAN) in Colab notebook for side-by-side comparison.
- Optimize hyperparameters and improve training stability.
- Add diverse evaluation metrics: FID, Inception Score (IS), Precision/Recall, and SSIM.
- Well-structured code for easy model switching.
- Comparative performance plots for quantitative metrics and generated image samples.
Ideal Skills and Experience:
- Proficiency in PyTorch and GANs.
- Strong background in hyperparameter optimization and stability improvement.
- Experience with evaluation metrics for generative models.
- Familiarity with Google Colab and modular code design.
Looking for a candidate who can deliver high-quality, diverse outputs and ensure a robust training process.
Key Requirements:
- Using multiple GAN architectures (WGAN-GP, DCGAN, StyleGAN, Conditional GAN) in Colab notebook for side-by-side comparison.
- Optimize hyperparameters and improve training stability.
- Add diverse evaluation metrics: FID, Inception Score (IS), Precision/Recall, and SSIM.
- Well-structured code for easy model switching.
- Comparative performance plots for quantitative metrics and generated image samples.
Ideal Skills and Experience:
- Proficiency in PyTorch and GANs.
- Strong background in hyperparameter optimization and stability improvement.
- Experience with evaluation metrics for generative models.
- Familiarity with Google Colab and modular code design.
Looking for a candidate who can deliver high-quality, diverse outputs and ensure a robust training process.