/image-sift-database

Primary LanguagePython

image-sift-database

Store your images SIFT descriptors using sqlite3.

Usage

See build.py for minimal usage:

from build_database import SIFTDatabase

SIFTDatabase().build((100, 100))

It

  1. creates a time-named database file
  2. CREATE TABLE sift (id INTEGER PRIMARY KEY, tag TEXT, descriptors BLOB)
  3. reads all the image files (ends with extension .jpg/.png) under images/
  4. resizes all the images to 100 * 100
  5. calculates the image descriptors
  6. INSERT INTO sift (tag, descriptors) VALUES (?, ?), [image_path.stem, descriptors.tobytes()]
  7. conn.commit()

API

from pathlib import Path

class SIFTDatabase:
    EXTS = [".jpg", ".png"]

    def __init__(
        self,
        images_path: Path = Path(os.getcwd()) / "images",
        output_file: str = f"sift_{int(time.time())}.db",
    ): ...

See build_database.py for details.

Notes

This repository's .gitignore is an "allow list":

**/*

!.git*
!/build_database.py
!/build.py
!/README.md

so you can safely put any files you need right after you clone this repository, and easily update the script with git pull. Have fun using!