our task, as noted above, is to build an image captioning model that will analyse input images and generate a short desc
Posted: Sat May 14, 2022 4:32 pm
our task, as noted above, is to build an image captioning model
that will analyse input images and generate a short
description.
An image captioning model in its simplest form is typically
composed of two parts:
Dataset:
The dataset you will use is the Flickr8k dataset, available from
a variety of online sources. It is composed of a training and a
testing sets.
Define your custom architecture that will receive images as
inputs and produce text as outputs. The goal is to generate the
most relevant and accurate image captioning as possible.
You must score your model (at minimum) with the scores BLEU-1,
BLEU-2, BLEU-3, BLEU-4.
However, there may be other metrics you wish to use to score
your model including visual
inspection and analysis among others.
that will analyse input images and generate a short
description.
An image captioning model in its simplest form is typically
composed of two parts:
Dataset:
The dataset you will use is the Flickr8k dataset, available from
a variety of online sources. It is composed of a training and a
testing sets.
Define your custom architecture that will receive images as
inputs and produce text as outputs. The goal is to generate the
most relevant and accurate image captioning as possible.
You must score your model (at minimum) with the scores BLEU-1,
BLEU-2, BLEU-3, BLEU-4.
However, there may be other metrics you wish to use to score
your model including visual
inspection and analysis among others.