GAN Skin Disease Image Generation

I'm seeking a deep learning specialist to execute an assignment focused on using a Generative Adversarial Network (GAN) for image generation. These aren't just any images though - I'm specifically after realistic skin disease photos. Ideally, the successful candidate will have: - Extensive experience in using GANs for image generation; - i have training data images and training labela and contour masks and also have the testing data - The ability to produce high quality, realistic images Attention to detail, reliability, and communication skills are paramount, as there will be a need for regular updates and iterations based on feedback. This is the task 3.1 Tasks 1. 2. 3.2 Implement a conditional GAN architecture conditioned on class labels to synthesize realistic images by taking unpaired sketches as input. Evaluate the model’s performance in terms of image quality, diversity, and realism. Analyze the effectiveness of the approach and identify areas for improvement. You need to report Frechet Inception Distance (FID) and Inception Score (IS). Experiments to perform: 1. Train cGAN and generate images according to the class labels. 2. Evaluate the generated images (at least 10 for each class) by considering them as test images and report the accuracy. Compare the accuracy with original test set images of the dataset. For this, you can train any classifier of your choice with the training data available to you. The dataset includes: • train data: training images. • train labels: training labels. • train sketch: unpaired training sketches. • test data: test images. • test labels: test labels. • test sketch: unpaired test sketches.