base on # Mirror > Hackable AI Powered Mirror on your laptop.  Mirror is a web app that constantly watches the realtime video feed from the webcam and responds with comments. 1. **100% Local and Private:** Try all kinds of ideas. Don't worry, everything happens on your laptop with NO Internet connection. 2. **FREE:** Since the AI model is running 100% on your machine, you can keep it running forever and experiment with different things. 3. **Hackable:** Simply by changing the prompt (or tweaking the code), you can easily repurpose Mirror to do different things. # How it works Watch the video of Mirror in action: [](https://www.youtube.com/watch?v=7Mx1W12Tvpw) 1. When you launch the app, the browser will ask you for webcam permission. 2. When you allow the webcam, it will start streaming the video to the AI ([Bakllava](https://huggingface.co/SkunkworksAI/BakLLaVA-1), running on [llama.cpp](https://github.com/ggerganov/llama.cpp)). 3. The AI will analyze the image and stream the response, which the frontend prints in realtime. # Usage When you launch the web UI, it will immediately start streaming responses from the AI based on the prompt: **"Describe a person in the image".** **You can edit this field** to let Mirror start streaming whatever you want  Some example prompts you can try: 1. What is this object I am holding? 2. What is the person doing? 3. Describe some notable events in the image. 4. How many people are in this picture? 5. Let me know if you see anything weird. # Install ## [RECOMMENDED] 1 Click Install Try the 1 click install using Pinokio: https://pinokio.computer/item?uri=https://github.com/cocktailpeanut/mirror > Make sure to use the latest version of Pinokio (0.1.49 and above)  > > Mirror has a lot of moving parts, so if you don't use the 1 Click Installer, it may take a lot of work: > > 1. Orchestration of multiple backends (llama.cpp server and the gradio webui server) > 2. Install pre-requisites, such as cmake, visual studio (windows), ffmpeg, etc. > If you want to install manually, go to the following section. ## Manual Install > Note that everything mentioned in this entire section is essentially what the 1 Click Installer does, automatically, and works on Mac, Windows, and Linux. So if you get stuck trying to run Mirror manually, try the 1 click install. ### 1. Clone this repository ``` git clone https://github.com/cocktailpeanut/mirror ``` ### 2. Clone llama.cpp ``` git clone https://github.com/ggerganov/llama.cpp ``` ### 3. Download AI Model Download the following bakllava model files to the `llama.cpp/models` folder - https://huggingface.co/mys/ggml_bakllava-1/resolve/main/ggml-model-q4_k.gguf - https://huggingface.co/mys/ggml_bakllava-1/resolve/main/mmproj-model-f16.gguf ### 4. Build llama.cpp ``` cd llama.cpp mkdir build cd build cmake .. cmake --build . --config Release ``` ### 5. Install requirements Create a venv and install rerquirements ``` python -m venv venv source venv/bin/activate pip install -r requirements.txt ``` ### 6. Install prerequisites Install FFMPEG: https://ffmpeg.org/download.html ### 7. Launch the llama.cpp server First start the llama.cpp server: #### Windows ``` cd llama.cpp\build\bin Release\server.exe -m ..\..\ggml-model-q4_k.gguf --mmproj ..\..\mmproj-model-f16.gguf -ngl 1 ``` #### Mac & Linux ``` cd llama.cpp\build\bin ./server -m ..\..\ggml-model-q4_k.gguf --mmproj ..\..\mmproj-model-f16.gguf -ngl 1 ``` #### 8. Launch the web UI First activate the environment: ``` source venv/bin/activate ``` Then run the app.py file ``` python app.py ``` # Credits 1. The backend code was inspired and adopted from [Realtime Bakllava](https://github.com/Fuzzy-Search/realtime-bakllava), which uses... 2. [Llama.cpp](https://github.com/ggerganov/llama.cpp) for the LLM Server. 3. [Bakllava](https://huggingface.co/SkunkworksAI/BakLLaVA-1) for the Multimodal AI model. 4. The Web UI was built with [gradio](https://www.gradio.app/). ", Assign "at most 3 tags" to the expected json: {"id":"4896","tags":[]} "only from the tags list I provide: [{"id":77,"name":"3d"},{"id":89,"name":"agent"},{"id":17,"name":"ai"},{"id":54,"name":"algorithm"},{"id":24,"name":"api"},{"id":44,"name":"authentication"},{"id":3,"name":"aws"},{"id":27,"name":"backend"},{"id":60,"name":"benchmark"},{"id":72,"name":"best-practices"},{"id":39,"name":"bitcoin"},{"id":37,"name":"blockchain"},{"id":1,"name":"blog"},{"id":45,"name":"bundler"},{"id":58,"name":"cache"},{"id":21,"name":"chat"},{"id":49,"name":"cicd"},{"id":4,"name":"cli"},{"id":64,"name":"cloud-native"},{"id":48,"name":"cms"},{"id":61,"name":"compiler"},{"id":68,"name":"containerization"},{"id":92,"name":"crm"},{"id":34,"name":"data"},{"id":47,"name":"database"},{"id":8,"name":"declarative-gui "},{"id":9,"name":"deploy-tool"},{"id":53,"name":"desktop-app"},{"id":6,"name":"dev-exp-lib"},{"id":59,"name":"dev-tool"},{"id":13,"name":"ecommerce"},{"id":26,"name":"editor"},{"id":66,"name":"emulator"},{"id":62,"name":"filesystem"},{"id":80,"name":"finance"},{"id":15,"name":"firmware"},{"id":73,"name":"for-fun"},{"id":2,"name":"framework"},{"id":11,"name":"frontend"},{"id":22,"name":"game"},{"id":81,"name":"game-engine "},{"id":23,"name":"graphql"},{"id":84,"name":"gui"},{"id":91,"name":"http"},{"id":5,"name":"http-client"},{"id":51,"name":"iac"},{"id":30,"name":"ide"},{"id":78,"name":"iot"},{"id":40,"name":"json"},{"id":83,"name":"julian"},{"id":38,"name":"k8s"},{"id":31,"name":"language"},{"id":10,"name":"learning-resource"},{"id":33,"name":"lib"},{"id":41,"name":"linter"},{"id":28,"name":"lms"},{"id":16,"name":"logging"},{"id":76,"name":"low-code"},{"id":90,"name":"message-queue"},{"id":42,"name":"mobile-app"},{"id":18,"name":"monitoring"},{"id":36,"name":"networking"},{"id":7,"name":"node-version"},{"id":55,"name":"nosql"},{"id":57,"name":"observability"},{"id":46,"name":"orm"},{"id":52,"name":"os"},{"id":14,"name":"parser"},{"id":74,"name":"react"},{"id":82,"name":"real-time"},{"id":56,"name":"robot"},{"id":65,"name":"runtime"},{"id":32,"name":"sdk"},{"id":71,"name":"search"},{"id":63,"name":"secrets"},{"id":25,"name":"security"},{"id":85,"name":"server"},{"id":86,"name":"serverless"},{"id":70,"name":"storage"},{"id":75,"name":"system-design"},{"id":79,"name":"terminal"},{"id":29,"name":"testing"},{"id":12,"name":"ui"},{"id":50,"name":"ux"},{"id":88,"name":"video"},{"id":20,"name":"web-app"},{"id":35,"name":"web-server"},{"id":43,"name":"webassembly"},{"id":69,"name":"workflow"},{"id":87,"name":"yaml"}]" returns me the "expected json"