Skip to content
Dataset Capture — Hakimi Sa’ari

Dataset Capture

Capture images, annotate bounding boxes, and export labeled datasets for computer vision.

Computer vision tooling · Personal project by Hakimi Sa’ari

Dataset Capture annotation editor showing a red chilli labeled Ripe chilli and the Update labels control.
Collect and label a ripe chilli for an agricultural dataset. Public sample photo with a manual demo box.Sample photo: Psychonaught / Wikimedia Commons · Public domain.

Open screenshot in the desktop Image Viewer

About this project

A mobile-friendly web app for moving from image collection to an annotated training dataset. It supports camera capture and bulk uploads, image browsing, explicit annotation saving, and YOLO ZIP export, alongside staff and administrator workflows.

What I built

  • Built camera capture, bulk image uploads, dataset browsing, and bounding-box annotation workflows.
  • Connected class labels and saved annotations to YOLO dataset exports.
  • Implemented staff/admin permissions, collection goals, filters, and dataset administration.
  • Supported Cloudflare storage and deployment, with a separate VPS runtime using SQLite and private filesystem images.

Tools

Next.js · React · TypeScript · Tailwind CSS · Cloudflare Workers · D1 · R2 · SQLite

Source repository is private.

All personal projects · Open on the desktop