Comparative analysis on cross-modal information retrieval: A review

Human beings experience life through a spectrum of modes such as vision, taste, hearing, smell, and touch. These multiple are integrated for information processing in our brain using complex network neuron connections. Likewise artificial intelligence to mimic the human way learning evolve into next generation, it should elucidate multi-modal fusion efficiently. Modality is channel that conveys about an object or event image, text, video, audio. A research problem said be when incorporates from more than single modality. Multi-modal systems involve one mode data inquired any (same varying) modality outcome whereas cross-modal system strictly retrieves dissimilar As input–output queries belong diverse modal families, their coherent comparison still open challenge with primitive forms subjective definition content similarity. Numerous techniques have been proposed by researchers handle this issue reduce semantic gap retrieval among different modalities. This paper focuses on comparative analysis various works field retrieval. Comparative several representations results state-of-the-art methods applied benchmark datasets also discussed. In end, issues presented enable better understanding present scenario identify future directions.