Repository files navigation

PiSAR: Pipeline for Aerial Search and Rescue AI-powered visual detection pipeline for aerial search operations

Try PiSAR online:

• Python 3.12+
• pip3 (Python package installer)
• (Optional) CUDA-enabled GPU

1. Clone the repository

   git clone https://github.com/eadali/PiSAR.gitcd PiSAR

2. (Recommended) Create a virtual environment

   python3 -m venv pisarsource pisar/bin/activate

3. Install dependencies

   • CPU only:

     pip3 install -r requirements.txt

   • GPU (CUDA) support:

     pip3 install -r requirements-cuda.txt

4. Verify installation

   python3 -c"import torch; print(torch.cuda.is_available())"

Seerequirements.txt andrequirements-cuda.txt for details.

To process an image, video, or camera stream, use the following commands:

python3 demo.py config/yolo8n-bytetrack-cpu.yaml --onnx-path downloads/yolo8n-416.onnx --video downloads/forest.mp4

python3 demo.py config/yolo8n-bytetrack-cuda.yaml --onnx-path downloads/yolo8n-416.onnx --video downloads/forest.mp4

Note:

• You must provide exactly one of--image,--video, or--camid.
• Theconfig argument is a positional argument (no--config).

Special thanks to theStephanST for providing the WALDO30 model used in this project.