ArchestackBest of Open-Source AI: List on GitHub ↗ (opens in a new tab)

#3737th of 60 in RAG, knowledge and search

sentrysearch

Semantic search over video footage that returns trimmed clips from text or image queries

Repository on GitHub ↗ (opens in a new tab)
Stars
4.6k
License
Apache-2.0
Last commit
Sep 2026
Language
Python

Missed the latest GitHub check; these facts may be older.

Overview

SentrySearch splits videos into overlapping chunks, embeds them with Gemini, Alibaba DashScope, LiteLLM or a local Qwen3-VL-Embedding model, and stores vectors in a local ChromaDB. Text or image queries are matched against the chunks, and the top match is trimmed from the source file with ffmpeg. It also offers anomaly-based highlights, VLM reranking, and MLX support on Apple Silicon.

Who it is for: Dashcam and security-footage users who want to search video by description

Strengths

  • Several backends: Gemini, qwen-cloud, LiteLLM, local PyTorch, and MLX for Apple Silicon
  • Fully local option keeps footage on your machine; vectors stored in local ChromaDB
  • Auto-trims matching clips; supports image queries, reranking and anomaly highlights
  • Backend and model auto-detected from the index; separate indexes avoid mixing embeddings

Weaknesses

  • Default Gemini backend sends video chunks to a cloud API and costs money
  • Requires Python 3.11 or 3.12; PyTorch wheels block 3.13+
  • Local backend is impractical on CPU-only or Intel Macs
  • MLX backend does not support --rerank; LiteLLM backend also lacks it

What it needs

  • GPU optional
  • Needs ffmpeg, ChromaDB, uv, Gemini API (default backend)
  • Models: Gemini Embedding, Qwen3-VL-Embedding 2B/8B, qwen3-vl-embedding (DashScope), LiteLLM gateways, MLX 4-bit

Also in RAG, knowledge and search

See all 60
Also in RAG, knowledge and search
RankProjectScore
1RAGFlowRAG engine with deep document parsing, template chunking and agentic retrieval Live demo ↗ (opens in a new tab)94.1k starsApache-2.090 out of 100
2Open NotebookSelf-hosted NotebookLM alternative for research notebooks, chat, and podcasts41.9k starsMIT90 out of 100
3LightRAGGraph-based RAG framework with a REST server and WebUI40.7k starsMIT90 out of 100
#4deeptutorTutoring workspace with knowledge bases, guided learning, memory and quizzes58.6k stars+17,457 (+42%) in 30 daysApache-2.089 out of 100
#5MaxKBEnterprise knowledge-base agent platform with RAG, workflows and MCP tools23.2k starsGPL-3.089 out of 100