Image Retrieval (IR) is one of the most exciting and fastest-growing research areas in the field of multimedia technology. We present here a highlight of recent research for IR. Some trends and probable future research directions are presented. We expose the major problems that we have recognized: the lack of a good measurement of visual similarity, the little importance accorded to user interaction and feedback, and the neglect of spatial information. Answering these concerns, we describe the solutions implemented by recent IR systems. We also present the current image retrieval projects in our laboratory, which are motivated to a large extent by these same considerations.