chn-lee-yumi/MaterialSearch

Semantic search. Search local photos and videos through natural language. AI语义搜索本地素材。以图搜图、查找本地素材、根据文字描述匹配画面、视频帧搜索、根据画面描述搜索视频。

What it solves

MaterialSearch allows users to search through their local collections of photos and videos using natural language descriptions or existing images, eliminating the need for manual tagging or complex folder structures to find specific visual media.

How it works

The system scans local directories for images and videos and uses a CLIP-based model (specifically chinese-clip-vit-base-patch16) to create searchable embeddings. It supports text-to-image, image-to-image, text-to-video, and image-to-video search, where video search identifies specific matching clips or segments based on the provided query.

Who it’s for

It is designed for individuals who manage large local libraries of visual assets and want a fast, AI-powered way to retrieve specific content without relying on traditional file names or metadata.

Highlights

  • Multimodal Search: Supports searching for images and videos using both text descriptions and image samples.
  • Video Segment Retrieval: Can find specific clips within videos rather than just the video file itself.
  • Hardware Flexibility: Works on CPU (amd64) and supports GPU acceleration for faster processing.
  • Easy Deployment: Available as a pre-packaged Windows bundle or a Docker image for streamlined setup.

Related

  • Project
    jd-opensource/JoyAI-VL-Interaction

    An 8B-scale vision-language interaction model and system that enables AI to proactively respond to real-time video streams in under a second without waiting for user prompts.

  • Project
    happier-dev/happier

    Happier is an open‑source, self‑hostable platform that lets you run, coordinate and steer multiple LLM‑based coding agents (Claude Code, Codex, OpenCode, etc.) across all your computers, servers and mobile devices. It provides a unified UI (desktop, web, iOS/Android) plus a CLI/daemon, supports parallel sessions, sub‑agents, Git workflow integration, voice input, and end‑to‑end encryption.

  • Project
    2U1/Qwen-VL-Series-Finetune

    A training toolkit for fine-tuning Qwen-VL series models (Qwen2-VL, Qwen2.5-VL, Qwen3-VL, Qwen3.5) supporting SFT, DPO, GRPO, and multimodal data.

  • Project
    android/androidify

    An open-source Android sample app that demonstrates how to build AI-driven experiences using the Gemini API, Imagen, and Jetpack Compose to rebuild the Android bot maker.