Show and Tell: A Neural Image Caption Generator

This code can be integrated with an existing project and captioning model can be load using object creation with loading the checkpoint just once.

• Pretrained Model with 5 Million checkpoints.
• Single instance loading.
• Removed Bazel-bin dependenct.
• Fixed vocab and checkpoints errors.

Install Required Packages

First ensure that you have installed the following required packages:

• Python 3.6 or greater
• TensorFlow Below 2.0(Ideal 1.15) (instructions)
• NumPy (instructions)
• Natural Language Toolkit (NLTK):
  • First install NLTK (instructions)
  • Then install the NLTK data package "punkt" (instructions)

Steps to run

Step 1: cd im2txt

Step 2: python im2txt.py

Results

Captions for image cycling.jpg:
  0) a person riding a bike on a city street (p=0.003317)
  1) a man riding a bike down a street . (p=0.000723)
  2) a person riding a bike on a street (p=0.000362)

Reference tensorflow model - Tensorflow: Show and Tell

If you find the repository useful, you can Star it as an appreciation.