Abstract: This work proposes an improved multi-modal matching algorithm based on object attention and tensor fusion (MOAT). The key improvements in proposed algorithm consists of two parts: (1) In ...