Skip to content

Files

Latest commit

 

History

History
22 lines (18 loc) · 853 Bytes

File metadata and controls

22 lines (18 loc) · 853 Bytes

Image Captioning Model

  • Uses pretrained ResNet50 model and Glove embeddings to caption any image
 Model ARCHITECTURE
 img feature -------->  MODEL --> Next word in sequence  ----
 partial sequence --->                                       |
   | _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ |
 Partial caption ----> RNN 
                           \
                            \ Feed forward network ----> predicted word,next
                            / ending with softmax        in the sequence of
                           /                               partial caption
               Image vector

Link for this model

Installation

pip install tensorflow
pip install keras