lancedb / yoloexplorer

YOLOExplorer : Iterate on your YOLO / CV datasets using SQL, Vector semantic search, and more within seconds

Developer-friendly, database for multimodal AI

---

LanceDB is an open-source database for vector-search built with persistent storage, which greatly simplifies retrevial, filtering and management of embeddings.

The key features of LanceDB include:

• Production-scale vector search with no servers to manage.

• Store, query and filter vectors, metadata and multi-modal data (text, images, videos, point clouds, and more).

• Support for vector similarity search, full-text search and SQL.

• Native Python and Javascript/Typescript support.

• Zero-copy, automatic versioning, manage versions of your data without needing extra infrastructure.

• GPU support in building vector index(*).

• Ecosystem integrations with LangChain 🦜️🔗, LlamaIndex 🦙, Apache-Arrow, Pandas, Polars, DuckDB and more on the way.

LanceDB's core is written in Rust 🦀 and is built using Lance, an open-source columnar format designed for performant ML workloads.

Quick Start

Javascript

npm install vectordb

const lancedb = require('vectordb');
const db = await lancedb.connect('data/sample-lancedb');

const table = await db.createTable({
  name: 'vectors',
  data:  [
    { id: 1, vector: [0.1, 0.2], item: "foo", price: 10 },
    { id: 2, vector: [1.1, 1.2], item: "bar", price: 50 }
  ]
})

const query = table.search([0.1, 0.3]).limit(2);
const results = await query.execute();

// You can also search for rows by specific criteria without involving a vector search.
const rowsByCriteria = await table.search(undefined).where("price >= 10").execute();

Python

pip install lancedb

import lancedb

uri = "data/sample-lancedb"
db = lancedb.connect(uri)
table = db.create_table("my_table",
                         data=[{"vector": [3.1, 4.1], "item": "foo", "price": 10.0},
                               {"vector": [5.9, 26.5], "item": "bar", "price": 20.0}])
result = table.search([100, 100]).limit(2).to_pandas()

Blogs, Tutorials & Videos

• 📈 2000x better performance with Lance over Parquet
• 🤖 Build a question and answer bot with LanceDB

yoloexplorer's Issues

Support for outer dataset formats?

Hi 👋🏻 Looks super interesting. Do you plan to expand to other dataset formats?

Support batch inference

How to use own dataset?

Hi there! Thanks for the great work!

If we want to use our own dataset, what should be included in the dataset? I mean, the directory format, labels, etc. Considering we have a dataset of images for a park, should we also have the labels for these images?

Hope this is not a stupid problem : )

Best regards,
Zijia

Roadmap, Scope and API design

Scope

Roadmap to V1:

• Dataset management - Simple ways to double down on specific parts of the dataset

Searching

• Using SQL

• Using Semantic search

• Support multiple datasets, i.e, allow semantic search across datasets (Via staging area)

• Using natural Language / AI search - Example- "Show me all images where a person is standing next to a car and the sample should be recorded before 2015"

• Allow enriching datasets by adding/removing samples from various sources, including other datasets ( secondary datasets of data lakes)

• Allow Various dataset formats (Users can load datasets from any format including 3rd party sources like RF, DagsHub, etc.)

• Allow loading tables directly - local, or cloud based ( s3 etc.)

• Allow various tasks:

• Detection
• Segmentation
• Classification
• ImgFolder structure used for generative AI or diffusion model samples

API Design:

• GUI/ Dashboard - TODO
• Pythonic/ Notebook - TODO

Contribution

Hello @AyushExel :wave, I want to contribute to this repo. Is there any contribution guideline or list of todos available?

Thank you.

Support multiple datasets

Add top level function to find duplicate imgs

Use recordbatch iterator when creating tables

Currently the table is built using the entire dataset in one go which'll probably fail for massive datasets due to memory constraints. Switch to iterator for initialization of LanceDB tables.

Add embedding visualization support

Create Dashboard

Allows users to plot images, analytics and embeddings in interactive browser plots. the current solution (matplotlib) isn't scalable.
The best candidate that I found was plotly dash(https://dash.plotly.com/) but there might be more so open to suggestions.
The idea is avoid building the UI from scratch using js/node as it'll make installation complex and development slow.

Make image storage optional to speed up table creation and improve memory footprint