A Hybridized Deep Learning Method for Bengali Image Captioning

An omnipresent challenging research topic in com-puter vision is the generation of captions from an input image. Previously, numerous experiments have been conducted on image captioning English but caption Bengali still sparse and need more refining. Only a few papers till now worked Bengali. Hence, we proffer standard strategy for two different sizes Flickr8k dataset BanglaLekha which only publicly available captioning. Afterward, our model were compared with generated by other researchers using architectures. Additionally, employed hybrid approach based InceptionResnetV2 or Xception as Convolution Neural Network Bidirectional Long Short-Term Memory Gated Recurrent Unit datasets. Furthermore, combination word embedding was also adapted. Lastly, performance evaluated Bilingual Evaluation Understudy proved that proposed indeed performed better consisting 4000 images dataset.