Back: Catalog

sentrysearch

SentrySearch is an open source tool based on Python, which allows you to perform semantic search on video recordings. By dissecting the videos, Gemini Embedding vectors through Qwen3-VL or cloud services and stores them on ChromaDB. Matches text or visual queries to export related video clips by cropping them out.

SE

sentrysearch

New65

ssrajadh · Search Engines

STARS4.6K
FORKS437
COMMITS0
CONTRIBUTORS0
VERSION—
UPDATED12 days ago
PythonVideo SearchGemini APIChromaDBApache-2.0

Common commercial products in this category: Algolia, Elastic Cloud, Azure AI Search

What can it do?
  • You can search video recordings by typing a text query and take the matching cross section as a clip.
  • You can search the video database using visuals.
  • With local backend or MLX support, you can trade on Apple Silicon without the need for a API key.
Heads up
  • • No published release
User reviews
  • ↓ NegativeGitHub Issue

    Broader application

    “## Problem What are you trying to do that SentrySearch doesn't support well today? Concrete example >I am trying to use sentrysearch to go over my football recordings for goals and saves. But it didn't work at all. I wonder if the model used in analyzing footage is about…”

  • ↓ NegativeGitHub Issue

    Bug: uv tool install on Windows defaults to CPU-only Torch (No GPU detection for local backend)

    “When following the README instructions to install SentrySearch as a global tool using uv tool install ".[local]", the installation defaults to the CPU-only version of PyTorch on Windows. This occurs because uv resolves the standard PyPI index, which does not include CUDA-enabled…”

GitHub

My note

Alternative pages

Algolia alternativesElastic Cloud alternativesAzure AI Search alternatives

More in: Search Engines