Image Search
What will I learn?
You will learn how to store brain scan images as vector embeddings in KDB.AI, then how to search this dataset to instantly retrieve similar images.
Agenda
In this tutorial you will walk through the process of storing images into KDB.AI using a pretrained neural network. You will then use nearest neighbor search capability to find and compare MRI Scans based on their Euclidean distance.
You will begin by loading the public dataset into your environment and then create embeddings using the ResNet-50 model. From here you will visualize the results and add the embeddings into kdb.ai.
Finally, you will perform a series of image similarity searches to test your embeddings.
Who is this suitable for?
Aimed at complete beginners to KDB.AI. No prior experience required but python knowledge will make it easier to follow along.
To access your own KDB.AI instance as shown in the video simply click the blue button above or else signup here.
The jupyter notebook shown in the video is available to download here. Note, you will need to install the requirements.txt file in the repository to make sure yo have all the relevent Python packages needed to run the code.