Light

pocinnovation / spacevector Goto Github PK

View Code? Open in Web Editor NEW

7.0 1.0 0.0 2.43 GB

SpaceVector is a platform for semantic search on satellite images using state of the art AI. It aims to support the use of satellite images.

Home Page: https://www.poc-innovation.fr/

License: Apache License 2.0

Dockerfile 8.52% Python 25.78% HTML 4.68% TypeScript 60.51% Shell 0.51%

ai pytorch satellite-imagery vision openai-clip vector-database

spacevector's Introduction

SpaceVector 🚀

SpaceVector is a platform for semantic search on satellite images using state of the art AI. It aims to support the use of satellite images.

How does it work?

How to SpaceVector works ?

When you make a query, the system embed you text with OpenAI Clip text encoder. Then, it search the most similar images in the vector database. The images are embed with OpenAI Clip image encoder. The system return the most similar images.

The majority images come from the EuroSat dataset. The dataset contains 27,000 labeled images of 10 classes. The images are 64x64 pixels and are in the RGB format. The classes are:

Annual Crop
Forest
Herbaceous Vegetation
Highway
Industrial
Pasture
Permanent Crop
Residential
River
Sea/Lake

Getting Started

Installation

For this project, you need to have Python installed on your computer.

You need docker and docker-compose to run the project.

Quickstart

For running the project, you need to run the following command:

docker-compose up

Usage

After you launch the docker compose, you can access the frontend on http://localhost:3000 and the backend on http://localhost:8000

Get involved

You're invited to join this project ! Check out the contributing guide.

If you're interested in how the project is organized at a higher level, please contact the current project manager.

Our PoC team ❤️

Developer

_{Clément Loeuillet}

Manager

_{Mikaël Vallenet}

Organization

🚀 Don't hesitate to follow us on our different networks, and put a star 🌟 on PoC's repositories

Made with ❤️ by PoC

spacevector's People

Contributors

Stargazers

Watchers

spacevector's Issues

[Feature]: Open Source Skills development

Description

Develop your Open Source skills 🎉

Expected behaviour

Here is what you have to do :

Environment

/

Proposed solution

/

[Feature]: Discover how implement OpenAI Clip in application

Description

The goal is to explore many ways to use OpenAI Clip in an application

Expected behaviour

The team have to :

Look this project which use OpenAI Clip
Check how implement Clip with HuggingFace transformers library
Check for others methods
Test in a simple file to use OpenAI clip

Environment

/

Proposed solution

/

[Feature]: Implement, deploy and fill the vector database

Description

You should implement the vector database, and deploy it.

Expected behaviour

The team have to :

Implement the vector database, you could check this
Deploy the database in local for now
Fill the database with sat images

Environment

/

Proposed solution

/

[Feature]: Learn about PyTorch & Neural Networks

Description

Complete the POC & 42AI technical exercise to gain skills in neural networks.

Expected behaviour

The jupyter notebook must be completed and documented on the methods used.
The students must be able to explain the basic functioning of a neural network.

The exercise is successful
The Notebook is documented
Visualization plots on loss and accuracy have been developed

Environment

/

Proposed solution

/

[Feature]: Learn about OpenAI Clip and understand its usefulness

Description

Students will discover OpenAI Clip 🧠

Expected behaviour

Team have to:

Look this article about Clip
Look this video about Clip
Write few sentences about what is Clip in the README.MD

Environment

/

Proposed solution

/

[Feature]: Backend

Description

The team must do:

Creation of a backend including the current application to return the desired images
Dockerization of the backend

Expected behaviour

/

Environment

/

Proposed solution

No response

[Feature]: Make a schema for the application

Description

You have to build the first step of the application : The vector database.
Since the application is complex, you should draw a schema of the application.

Expected behaviour

The team have to:

Draw a schema
Put the schema in README.MD
Explain the schema in README.MD

Environment

/

Proposed solution

/

[Feature]: Frontend

Description

The team must do:

Creation of a model for the application
Creation of the application with the model
Dockerization of the frontend

Expected behaviour

/

Environment

/

Proposed solution

No response

[Feature]: Devops

Description

The team must do:

Create a docker-compose global
Create script to fill the database at creation of database

Expected behaviour

/

Environment

/

Proposed solution

No response

[Feature]: Fill the database with many datasets

Description

For better results, it is now necessary to fill the database with many datasets.

Expected behaviour

The team hate to:

Find datasets
Add them to the database

Environment

/

Proposed solution

/

[Feature]: Read the papers and write a resume

Description

The team will read the paper about OpenAI Clip and understand it 🚀

Expected behaviour

The team have to:

Read the paper
Look this video
Write a document in french which resume OpenAI Clip
Students must be able to explain the process of OpenAI Clip

Environment

/

Proposed solution

/

[Feature]: Questions & Answers about the project

Description

Prepare questions about the project and store answers 🚀

Expected behaviour

Team have to :

Prepare questions about the project
Write answers on notion

Environment

/

Proposed solution

/

[Feature]: Increase general project documentation

Description

The team must do:

Explain the overall functioning of the project
Add schema in README.md
Explain the difficulties encountered and the failures

Expected behaviour

/

Environment

/

Proposed solution

No response

[Feature]: Read Stanford course about Computer Vision

Description

I want the team to read the Computer Vision CS231n course. 📑

Expected behaviour

The course must be read and understood. The team must be able to answer my questions on the subject.

The course is read
Team is able to answer my questions

Environment

/

Proposed solution

/

[Feature]: Add file path in database

Description

Add a field named "path" in database.

Expected behaviour

Add field
Clear database
Fill database with new data

Environment

/

Proposed solution

No response

[Feature]: Implement an AI Pipleline for CIFAR10-Dataset

Description

You have the CIFAR-10 dataset which contains 60,000 images, and 10 different labels.

Implement an application that uses computer vision. 🏅

Expected behaviour

Your application should:

getting started with python3
take as parameter an argument which is a folder
perform start-up training
display the evolution of the loss and accuracy during training
Make predictions once trained on all the images that the folder passed in parameter contains.
Write a predictions output file that maps each image name to its label
Make an application Gradio

Bonus ideas

Propose to the user to choose his hyperparameters
implement a graphical interface using Streamlit or Gradio
Integrate an AI stack like: Hydra, Optuna & DVC

Environment

/

Proposed solution

/

[Feature]: Implement OpenAI Clip in an application usable

Description

Create an application wich use OpenAI Clip 🚀

Expected behaviour

The team have to :

Create an application which use OpenAI Clip
The application should be easily usable

Environment

/

Proposed solution

/

[Feature]: Data Science, Logistic Regression, Neural Network & DecisionTree skills Skils

Description

The objective is to complete the various workshops on data science, logistic regression and decision trees.
Students will be able to explain the different concepts related to the workshops.

Expected behaviour

Workshop Data Science
Workshop Logistic Regression
Workshop DecisionTree
Workshop Neural Network

Environment

/

Proposed solution

/

[Feature]: Retrieve an image from the database

Description

After querying the database and retrieving the image vector, we now need to retrieve the image to show it.

Expected behaviour

The team have to :

Convert a vector to an image

Environment

/

Proposed solution

/

Application with OpenAI Clip

Description

Description
The goal is to create an application with Streamlit on Hugging face that uses OpenAI Clip

Expected behaviour
The team have to :

Use Streamlit
To be able to obtain images from a text

Expected behaviour

/

Environment

Hugging Face

Proposed solution

No response

Recommend Projects

React

A declarative, efficient, and flexible JavaScript library for building user interfaces.
Vue.js

🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
Typescript

TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
TensorFlow

An Open Source Machine Learning Framework for Everyone
Django

The Web framework for perfectionists with deadlines.
Laravel

A PHP framework for web artisans
D3

Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

javascript

JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
web

Some thing interesting about web. New door for the world.
server

A server is a program made to process requests and deliver data to clients.
Machine learning

Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Visualization

Some thing interesting about visualization, use data art
Game

Some thing interesting about game, make everyone happy.

Recommend Org

Facebook

We are working to build community through open source technology. NB: members must have two-factor auth.
Microsoft

Open source projects and samples from Microsoft.
Google

Google ❤️ Open Source for everyone.
Alibaba

Alibaba Open Source for everyone
D3

Data-Driven Documents codes.
Tencent

China tencent open source team.