# ngxson/smolvlm-realtime-webcam

Real-time webcam demo with SmolVLM and llama.cpp server

Repository: https://github.com/ngxson/smolvlm-realtime-webcam
Canonical: https://ross.abutalabs.com/products/smolvlm-realtime-webcam
Homepage: https://github.ngxson.com/smolvlm-realtime-webcam/
Language: HTML
License: NOASSERTION
License Family: other
Last push: 2025-05-12T17:24:39+00:00

## Health v2 (maintenance only)
Score: 29/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 21, release rhythm 35, longevity 34
- inputs: {"age_days": 478, "days_push": 478, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 5571, forks 894 (observed 2026-08-28T04:09:22.490199+00:00)

## What it is
A browser-based demo that streams webcam frames to a llama.cpp server running SmolVLM 500M for real-time object detection and scene description. It consists of a single HTML page that periodically sends camera snapshots to the local vision model API.

## Use cases
- run real-time object detection with a small vision model on my webcam
- demo multimodal vision-language models locally with llama.cpp
- describe what my camera sees in real time using SmolVLM
- experiment with local VLM inference from the browser
- build a webcam AI assistant that returns JSON descriptions

## When to choose
- you want a lightweight, local real-time webcam vision demo with no cloud dependency
- you already run llama.cpp and want to try multimodal models
- you need a simple reference implementation for polling a VLM with camera frames

## When to avoid
- you need production-grade, high-accuracy object detection
- you want a full application with recording, alerts, or multi-camera support
- you cannot run a local llama.cpp server with adequate GPU/CPU performance

## Facets
- artifact type: application
- maturity: active
- function: computer-vision, llm-inference, image-processing
- domain: artificial-intelligence, computer-vision, large-language-models
- platform: browser, cross-platform
- tags: smolvlm, llama-cpp, webcam, real-time, demo, multimodal, gguf, web-server

## Member repositories
- ngxson/smolvlm-realtime-webcam (main) score 29

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:22.490199+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:55:35.468398+00:00, confidence not recorded.
  - readme: https://github.com/ngxson/smolvlm-realtime-webcam (fetched 2026-08-28T04:09:22.490199+00:00, sha e7153f86d0f8)
  - homepage: https://github.ngxson.com/smolvlm-realtime-webcam/ (fetched 2026-08-29T08:50:37.696733+00:00, sha c2cb4517e017)
- Data as of 2026-08-30T08:39:29.467469+00:00.