Python Image Processing Cookbook

How it works...

The build_graph_from_config() function builds the inference graph from a configuration object. It takes a model config object containing the configuration for building the model and a checkpoint path containing the checkpoint file. It returns a function that can be used to load the model variables from the checkpoint file.

The beam_search() function runs beam search caption generation on a single image. It accepts a TensorFlow session object and a CNN-encoded image string as input and returns a list of captions sorted by descending score.

See caption_generator.py for a description of the available beam search parameters.