Skip to content
Video Annotation Studio — Hakimi Sa’ari

Video Annotation Studio

Extract useful frames from video, generate YOLO labels, and review annotations.

Computer vision tooling · Personal project by Hakimi Sa’ari

Video Annotation Studio reviewing a shuttle launch frame with a bounding box, class controls, an annotation list, and navigation through five frames.
Review shuttle labels across five frames extracted from launch footage. Manual demo boxes; frame navigation and correction controls are shown.Footage: NASA, STS-129 launch · Public domain.

Open screenshot in the desktop Image Viewer

About this project

A web-based dataset-preparation tool that connects video extraction, model-assisted annotation, and manual review. Users upload a YOLO model, choose which detection classes to extract, and refine the resulting frames and labels.

What I built

  • Built model upload and video extraction with class filtering and frame-skip controls.
  • Used YOLO predictions to prepare annotations for extracted frames.
  • Added class mapping and manual editing of bounding boxes and labels.
  • Created review workflows for correcting or removing frames and annotations.

Tools

Python · Flask · YOLO · JavaScript · HTML & CSS

View public source

All personal projects · Open on the desktop