whereismyai
SentrySearch
SentrySearch is a semantic video search tool that matches text or image queries to trimmed clips.
SentrySearch performs semantic search over video footage by splitting files into overlapping chunks and embedding them using Google Gemini, Alibaba DashScope, LiteLLM, or local models. Users can search using text queries, reference images, or generate automated anomaly highlights using methods like KNN, centroid, or Local Outlier Factor. It stores vectors in a local ChromaDB database and automatically trims top matches into saved clips using FFmpeg. The tool is designed for anyone needing to locate specific moments in video datasets without manual scrubbing. It supports features like deduplication, VLM reranking, and specialized Apple Silicon MLX acceleration.
SentrySearch at a glance
- Best for: Software engineer
- Runs on: macOS, Windows, Linux, API, self-hosted, open source
- Pricing: Free
- Released: March 2026
- Category: AI search engine
- Website: github.com/ssrajadh/sentrysearch
Key features
- Index video footage into local ChromaDB
- Search video using text queries
- Search video using reference images
- Generate automated anomaly highlights
- Rerank results with vision-language models
- Automatically trim matches into clips
- Run locally using MLX or Qwen models
Use it for
- Finding specific visual events in video archives
- Searching security camera footage using text prompts
- Generating automated highlight clips from raw footage
- Querying video files using reference images
Alternatives to SentrySearch
- SCM — Runs 100% locally on Apple Silicon Mac with no cloud uploads.
- Examples — Runs entirely in your browser for fast semantic image search.
- Hister — Protects privacy with a self-hosted search index for your pages and files.
- Meilisearch — It returns answers in less than 50 milliseconds with out-of-the-box relevancy.
- XERJ — It answers agent questions from an index instead of wasting tokens on context windows.
- Orcah Studio — It indexes video files locally on your computer with no cloud dependency.
- V-Modal — Find any video frame or audio segment using text descriptions
- SeekStorm — It delivers sub-millisecond lexical and vector search performance in Rust.
- DigUp — It searches local files by content and meaning without sending data to the cloud.
- Vespa — Combines vector database capabilities with distributed machine-learned model inference.
- Gsearch — Runs a real browser over MCP to fetch genuine Google results with zero API keys.
- PixelRAG — Preserves visual structure like tables and charts that text parsers miss.
All SentrySearch alternatives