Course videos hide structured learning goals inside transient slides that are slow to review manually across long playlists.
Video Goals Extractor
A local web app and CLI that processes YouTube videos and playlists, extracts slide frames, reads Arabic and English learning goals with OCR or vision AI, and exports structured reports.
CONTEXT & DELIVERY
The engineering story.
A media pipeline downloads permitted inputs, samples frames, detects useful slides, applies OCR or vision analysis, deduplicates results, and exports a reviewable report.
Pipeline design, local web interface, CLI workflow, media processing, OCR and vision integration, and structured export.
A focused automation tool that converts long-form educational video material into a concise, reviewable set of learning goals.
SYSTEM DESIGN
How the product works.
Media intake
The local workflow accepts supported YouTube videos and playlists for processing.
INPUTFrame sampling
FFmpeg extracts representative frames from the timeline.
FRAMESText and vision
OCR or vision analysis reads Arabic and English learning goals from useful slide frames.
OCRDeduplication
Repeated frames and repeated goals are reduced before export.
CLEANStructured report
The final result is organized for review instead of remaining buried inside video timelines.
REPORTENGINEERING DECISIONS
What keeps it trustworthy.
Processing stays under user control
The tool is designed as a local workflow rather than a hosted media collection service.
Automation keeps human review
Extracted goals remain visible and structured before they are reused.
One problem, one pipeline
The product stays narrow and optimizes the path from video slide to usable learning goal.
04 / TECHNOLOGY
Tools chosen for the job.
Claude VisionFastAPIFFmpegPythonTesseract OCRyt-dlp