From cameras to clips.
One step at a time.

10–20 minutes at the rink

Plan a stable, overlapping view

Place two cameras high enough to clear the glass and spectators. Aim one toward each half of the playing surface, with a generous overlap around center ice. Lock both cameras to the same resolution, frame rate, and color settings.

• Use sturdy mounts and external power for long tournament days.
• Start both recordings before warmup so calibration has clear frames.
• Keep camera positions fixed for the entire game.

Optional hardware

Build the remote and printable hardware

GoProHockeyMON pairs firmware for an ESP32-S3 1.8-inch AMOLED remote with a catalog of printable cases, cooling parts, mounts, and rink-side transport hardware.

git clone https://github.com/cjolivier01/GoProHockeyMON.git
cd GoProHockeyMON
make build
make upload-monitor

Put the GoPro in Bluetooth pairing mode. On the remote, choose Pair New. Once paired, the remote remembers the camera, can preview a snapshot, and can start recording from the side button.

make -C models3d
make -C models3d hockeymon-camera-case

After the game

Organize the camera files

Copy each camera’s MP4 chapters into a game folder. Both HockeyMON and HockeyMONStream understand this layout, including common GoPro and Insta360 chapter naming patterns.

By default, the tools look under $HOME/Videos/<game-id>. Set HM_GAME_DIR if your game library lives elsewhere.

Fast production path

Stitch and process with HockeyMONStream

HockeyMONStream is the high-performance C++ application. It requires NVIDIA DeepStream, CUDA, TensorRT, and the dependencies documented with the release. Installed commands and package paths retain the hstream name for compatibility.

From a source checkout, run the game end to end:

git clone https://github.com/cjolivier01/HockeyMONStream.git
cd HockeyMONStream
./perf
./run.sh --game-id=sharks-v-falcons -t=5

The final archive is published to the game folder as sharks-v-falcons-tracking_output-with-audio.mp4.

Flexible Python path

Track a panorama with HockeyMON

Use HockeyMON when you want the Python pipeline, model experimentation, or a simpler command against video that has already been stitched.

hmtrack \
  --input-video stitched_panorama.mp4 \
  --output tracking.mp4

For a configured game directory, hmtrack --game-id sharks-v-falcons runs the full stitch-and-track workflow. Configuration files can be layered with repeated --config arguments.

Team and season view

Bring schedules, stats, and clips together

HockeyMONWeb combines public league data with team-managed imports. Run it locally with SQLite for evaluation, or follow its deployment tools for a persistent MariaDB-backed installation.

git clone https://github.com/cjolivier01/HockeyMONWeb.git
cd HockeyMONWeb
python3 -m venv .venv
. .venv/bin/activate
python -m pip install -r requirements-dev.txt
cd tools/webapp && python3 manage.py runserver