Vision App

This project is a computer vision application designed to detect humans and cats from video streams. It uses the YOLOv8 model for object detection and can analyze images with an LM Studio server.

Features

• Real-time object detection: Detects objects in video streams using the YOLOv8 model.
• Motion detection: Only processes frames when motion is detected.
• LM Studio integration: Sends image snapshots to an LM Studio server for further analysis.
• Extensible architecture: Easily supports new video stream sources.
• Terminal UI: Includes a terminal-based user interface for monitoring.

Installation

1. Clone the repository:

   git clone https://github.com/SysDevJT/pidetect
   cd detect

2. Install dependencies:

   pip install -r requirements.txt

3. Download the YOLOv8 model:

   wget https://github.com/ultralytics/assets/releases/download/v0.0.0/yolov8n.pt

It wil most likely download itself.

Configuration

The application is configured through environment variables. You can set the following variables:

• RTSP_URL: The URL of the RTSP stream to connect to.
• USE_PICAM: Set to true to use a PiCamera instead of an RTSP stream.
• LMSTUDIO_URL: The URL of the LM Studio server.
• LMSTUDIO_MODEL: The name of the model to use in LM Studio.
• WEBHOOK_URL: The URL to send webhook notifications to.

Usage

To run the application, use the following command:

python3 detect_picam_rtsp.py

Examples

Using picam

USE_PICAM=1 python detect_picam_rtsp.py

Using rtsp:

RTSP_URL=rtsp://username:password@127.0.0.1:554 python3 detect_picam_rtsp.py

Use the credentials of the rtsp cam.

See the config.py if you want to use a default.