MediaPipe Object Detection Demo

This guide shows how to implement MediaPipe graph using OVMS.

Example usage of graph that accepts Mediapipe::ImageFrame as a input:

Prepare the repository

Clone the repository and enter mediapipe object_detection directory

git clone
cd model_server/demos/mediapipe/object_detection

Prepare models and the environment:

virtualenv .venv
. .venv/bin/activate
pip install -r requirements.txt

python --download_models

Run OpenVINO Model Server

docker run -d -v $PWD/ovms:/demo -p 9000:9000 openvino/model_server:latest --config_path /demo/config.json --port 9000

Run the client:

python --grpc_port 9000 --images ./input_images.txt
Start processing:
  Graph name: objectDetection
Iteration 0; Processing time: 41.05 ms; speed 24.36 fps
Iteration 1; Processing time: 25.04 ms; speed 39.93 fps
Iteration 2; Processing time: 29.88 ms; speed 33.46 fps
Iteration 3; Processing time: 26.61 ms; speed 37.59 fps

Received images with bounding boxes will be located in ./results directory.