Skip to content

Multimodal vector search of images and videos taken from trail cameras. Demonstrates how build multimodal AI search (text and image) using the Meta AI ImageBind model.

Notifications You must be signed in to change notification settings

redswimmer/trail-camera-search

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

14 Commits
 
 
 
 
 
 
 
 

Repository files navigation

Trail Camera Image and Video Search with Meta AI ImageBind

This example demonstrates how to build a multi-modal search (image and video) using the Meta AI ImageBind and the Weaviate vector database and was based off an example from Weaviate here.

Querying using one modality (e.g. text) will include results in all available modalities (e.g. images and video), as all objects will be encoded into a single vector space.

Here is a link to a demo video I recorded if you don't want to play with the code.

Weaviate Setup

The ImageBind model is only available with local Weaviate deployments with Docker or Kubernetes.

ImageBind is not supported with Weaviate Cloud Services (WCS).

Steps to deploy Weaviate locally with ImageBind

  1. Install Docker.

    If you are new to Docker Compose, here are instructions on how to install it.

  2. Run Weaviate+Bind with Docker Compose

    In the terminal, navigate to the root director of this project and locate the file docker-compose.yml and call:

    docker compose up
    

    Note #1 - the first time you run the command, Docker will download a ~6GB image.

    Note #2 – running this Docker image requires 12GB of RAM. If you're in Windows you'll need to adjust your .wslconfig to include the following:

    [wsl2]
    memory=12GB
    

About

Multimodal vector search of images and videos taken from trail cameras. Demonstrates how build multimodal AI search (text and image) using the Meta AI ImageBind model.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published