DOCSEARCHbyAlgolia
Submit Docs

smolvlm-realtime-webcam

repository·main·Indexed 26 days ago

https://github.com/ngxson/smolvlm-realtime-webcam

A demonstration project for real-time object detection using a webcam feed and the SmolVLM-500M-Instruct-GGUF model served via llama.cpp.

Tokens
158
Snippets
0
Records
2
Agent score
41%

What's inside smolvlm-realtime-webcam

  2. Setup the SmolVLM real-time camera demo

    main

    To run the demo, follow these steps:

    1. Install llama.cpp.
    2. Start the llama-server using the SmolVLM model. Use the -ngl 99 flag if you want to enable GPU acceleration (for NVidia, AMD, or Intel GPUs).
    3. Open the index.html file in your web browser.
    4. (Optional) Modify the instruction in the application (e.g., to request JSON output).
    5. Click the "Start" button in the browser to begin the demo.