Performing text-to-video retrieval with TensorFlow Hub
The applications of deep learning to videos are not limited to classification, categorization, or even generation. One of the biggest resources of neural networks is their internal representation of data features. The better a network is at a given task, the better their internal mathematical model is. We can take advantage of the inner workings of state-of-the-art models to build interesting applications on top of them.
In this recipe, we'll create a small search engine based on the embeddings produced by an S3D model, trained and ready to be used, which lives in TFHub.
Are you ready? Let's begin!
Getting ready
First, we must install OpenCV and TFHub, as follows:
$> pip install opencv-contrib-python tensorflow-hub
That's all we need, so let's start this recipe!
How to do it…
Perform the following steps to learn how to perform text-to-video retrieval using TFHub:
- The...
