Video frame extraction

Skill cxcscmu/SkillLearnBench/skills/b4-skill-creator-claude-sonnet-4-6/video-object-counting/video-frame-extraction

Extract key frames (I-frames) from video files using FFmpeg. Use this skill whenever the user needs to pull keyframes, thumbnails, or important frames from MP4, MKV, AVI, or other video formats. Outputs PNG image files named with zero-padded indices (e.g., keyframes_001.png).From its SKILL.md

Install

npx -y skills add cxcscmu/SkillLearnBench --skill video-frame-extraction

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

SKILL.md

1.4 KB, 312 tokens by cl100k_base, as published. Nobody here has run it

Video Frame Extraction with FFmpeg

Extract I-frames (keyframes) from a video file and save them as sequentially numbered PNG images.

Command

ffmpeg -i <input_video> -vf "select=eq(pict_type\,I)" -vsync vfr <output_dir>/keyframes_%03d.png

select=eq(pict_type\,I) — selects only I-frames (intra-coded keyframes)
-vsync vfr — variable frame rate output to avoid duplicates
%03d — zero-padded 3-digit index starting at 1 (e.g., 001, 002, ...)

Example

ffmpeg -i /root/super-mario.mp4 -vf "select=eq(pict_type\,I)" -vsync vfr /root/keyframes_%03d.png

Output files: /root/keyframes_001.png, /root/keyframes_002.png, etc.

Listing Extracted Frames

After extraction, list frames sorted by name to get timeline order:

import glob, os
frames = sorted(glob.glob('/root/keyframes_*.png'))

Notes

I-frames are the most information-rich frames and good representatives for scene analysis.
Frame indices start at 1, not 0.
If the video has no explicit I-frames, use -vf fps=1 to extract one frame per second instead.

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.