A Lightweight Generalizable Evaluation and Enhancement Framework for Generative Models and Generated Samples

While extensive research has been conducted on evalu-ating generative models, little research has been conducted on the quality assessment and enhancement of individual-generated samples. We propose a lightweight generaliz-able evaluation framework, designed to evaluate and en-hance the generative models and generated samples. Our framework trains a classifier-based dataset-specific model, enabling its application to unseen generative models and extending its compatibility with both deep learning and ef-ficient machine learning-based methods. We propose three novel evaluation metrics aiming at capturing distribution correlation, quality, and diversity of generated samples. These metrics collectively offer a more thorough performance evaluation of generative models compared to the Fréchet Inception Distance (FID). Our approach assigns individual quality scores to each generated sample for sample-level evaluation. This enables better sample mining and thereby improves the performance of generative mod-els by filtering out lower-quality generations. Extensive ex-periments across various datasets and generative models demonstrate the effectiveness and efficiency of the proposed method.