This document provides an overview of content-based image retrieval with relevance feedback using soft computing techniques. It discusses CBIR and the problems with semantic gaps between low-level features and high-level semantics. Relevance feedback is introduced as a technique to refine queries to reduce this gap, but it decreases system performance. The document then reviews related work applying machine learning methods like SVM and AdaBoost to relevance feedback. It also introduces soft computing methods like neural networks, genetic algorithms, and fuzzy clustering to improve retrieval efficiency and performance. Finally, it discusses measures like precision and recall for evaluating system performance.