Go2Rep

Go2Rep is a GUI for multi-GoPro HERO (tested on GoPro 11/13) video collection and processing (COHN or BLE+AP). It supports camera control, video sync, Theia classification, calibration, and report generation—built for markerless 3D motion capture using wireless GoPro workflows.

0. Installation

Installation via Anaconda

1. Click on the link to download the anaconda environment : (Link1)

2. Add Go2Rep to your Anaconda environment:

   • Navigate to:
     C:\ProgramData\anaconda3\envs\
   • Locate your environment folder (e.g., Go2Rep).
   • Ensure that Go2Rep is properly placed in:
     C:\ProgramData\anaconda3\envs\Go2Rep\Lib\site-packages\Go2Rep

3. Launch the GUI:

   • Open the terminal or Anaconda Prompt.
   • Navigate to the folder:

     cd C:\ProgramData\anaconda3\envs\Go2Rep\Lib\site-packages\Go2Rep

   • Activate Go2Rep environment:

     conda activate Go2Rep

   • Run the following command:

     python main_gui.py

Download the Example Project

If you want, you can download the Go2Rep example by clicking on this link:
Link1

After downloading, extract and place the contents into the following directory:
C:\ProgramData\anaconda3\envs\Go2Rep\Lib\site-packages\Go2Rep\Example_Go2Rep

---

Troubleshooting of Anaconda Installation

Problem 1

'python' is not recognized as an internal or external command

Cause:
Python is either not installed or its path is not added to the system’s PATH variable.

Solution:

Step 1: Check if Python is installed
Open CMD and type:

python --version

If you get 'python' is not recognized, proceed to the next step.

2. Install Python
   Download Python from https://www.python.org/downloads/
   During installation, make sure to check the box:

Add Python to PATH
before clicking Install Now.

3. Add Python to PATH manually (if needed)

• Open:
  Control Panel > System > Advanced system settings > Environment Variables
• Under System variables, select Path and click Edit.
• Click New and add the folder path where python.exe is installed, for example:

  C:\ProgramData\anaconda3
  python --version
  

4. Restart Command Prompt and verify

• Run:
  Control Panel > System > Advanced system settings > Environment Variables
• Under System variables, select Path and click Edit.
• Click New and add the folder path where python.exe is installed, for example:

  C:\ProgramData\anaconda3
  python --version
  

Problem 2:

nvcc compilation failed / Cannot find compiler 'cl.exe' in PATH

Cause:
CUDA’s nvcc compiler needs the Microsoft C++ compiler (cl.exe), which isn’t found in your system PATH.

Solution:

1. Install Visual Studio with C++ Build Tools
   Download Visual Studio from https://visualstudio.microsoft.com/
   During installation, select the workload:

Desktop development with C++

2. Activate Visual Studio environment variables
   Open a command prompt and activate your conda environment:

conda activate Go2Rep
cd C:\ProgramData\anaconda3\envs\Go2Rep\Lib\site-packages\Go2Rep

Then run the Visual Studio environment setup script:

"C:\Program Files\Microsoft Visual Studio\2022\Community\VC\Auxiliary\Build\vcvars64.bat"

This configures environment variables so cl.exe can be found.

3. Run your GUI script

   python main_gui.py

Prerequisites

This GUI has several system requirements, which are outlined in the following post: (https://github.com/ShabahangShayegan/Go2Rep/wiki#prerequisites)

1. Features

Here, we present the general functionality of the GUI. For more details about each function, please refer to the Wiki page.

1.0 Visual flow chart of Go2Rep

1.1 GoPro Control

Go2Rep provides a unified interface for controlling multiple or mono GoPro HERO cameras (tested on HERO11 and HERO13) over wireless connections.(via BLE,AP and COHN)
It handles the full workflow from discovery and connection to recording, previewing, downloading, and organizing videos.
The application automatically adapts its communication method based on the camera model, ensuring the correct protocol is used for reliable control and media transfer.

Key capabilities include:

• Camera discovery via Bluetooth
• Connection management (via BLE,AP and COHN)
• Recording controls (start/stop, FPS, resolution)
• Automatic media collection with proper handling of encrypted downloads
• Model-specific workflows to match protocol and provisioning requirements

---

Feature	GoPro 11 or older	GoPro 13 / 12
Protocols Used	BLE + WiFi AP	HTTPS via COHN
Certificate Provisioning	❌ Not Required	✅ Required
Preview & Streaming	Supported via WiFi AP	Supported via COHN*
Media Download	Basic via BLE + WiFi	Encrypted via HTTPS

*Not implemented so far

1.2. Synchronization

The synchronization feature aligns multiple videos recorded during the same trial to enable frame-accurate analysis across cameras. It supports two main methods:

1.2.1 Manual Synchronization

A GUI-based tool with CUDA acceleration that lets users manually select matching frames across videos. Videos are grouped by timestamped filenames and users pick a reference video, define a frame range, then visually select corresponding frames in other videos to calculate frame offsets. The output is a JSON file listing frame offsets per video.

Key points:

• Requires manual frame selection aided by CUDA video processing.
• Require videos contain a clear synchronization event.
• Outputs offsets for temporal alignment of all videos.

1.2.2 Timecode Synchronization

An automated method using embedded video metadata (creation_time and timecode) extracted via ffprobe. Videos are grouped by filename timestamps, and synchronization offsets are calculated by comparing each video’s start timecode to the earliest one.

Key points:

• Requires videos with accurate embedded timecode metadata (Only recommended for GoPro 12/13).
• Requires scanning the Precision Time QR codes or using time synchronization from GoPro Quik to obtain highly accurate timestamp alignment across cameras. This step ensures sub-frame synchronization precision by leveraging GoPro's Precision Time protocol documented at GoPro Precision Time.
• Outputs JSON and CSV files with frame offsets per video.
• May suffer from drift and requires careful setup (e.g., GPS off or GPS_SYNC=0).

1.3. Classification

Processes and synchronizes multi-camera trial videos using filename timestamps or an optional synchronization JSON file. The system groups videos into trials based on timing proximity, trims them to ensure frame alignment, and re-encodes them in the selected format (e.g., .mp4).

Important:
Theia requires all videos in a trial to have the same number of frames and to be re-encoded in a compatible format. This tool ensures that by trimming all videos in each trial to the shortest one, or using synchronization offsets (if provided) and further trimming them to the minimum valid range across all videos before re-encoding.

The input of this module is Folder containing raw videos coming from video storage directory and pptional synchronization JSON file. The output is Classified, synchronized, and re-encoded video sets per trial and camera, compatible with Theia requirements.

Keypoints

• Optional synchronization: Applies frame offsets from a synchronization JSON file to align videos temporally.
• Trimming:
  • Without a synchronization file, videos are trimmed to match the shortest video’s length in the trial.
  • With a synchronization file, trimming respects the provided frame offsets and defined start/end frame ranges, ensuring precise temporal alignment before further trimming all videos to the minimum valid frame range.
• Re-encoding: Videos are re-encoded in the selected output format to ensure compatibility with Theia.
• Output: Saves synchronized, trimmed, and re-encoded videos in trial- and camera-specific folders ready for Theia analysis.

1.4 Calibration

This module automates camera calibration (intrinsic and extrinsic parameters) essential for accurate 3D reconstruction and measurement from video data. It uses OpenCV and provides an interactive GUI for manual confirmation or selection of known 3D scene points, following a workflow similar to Pose2Sim

1.4.1 Calibration Workflow

• Inputs:

  • Videos or images showing at least six visible, known 3D scene points
  • config.toml with measured 3D points and calibration settings
  • Pre-computed intrinsic parameters file (Calib.toml) from Pose2Sim

• Process:

  1. User launches calibration from the GUI and selects input/output folders.
  2. The system attempts to detect known points automatically; users can confirm or manually select points via an interactive window.
  3. Manual controls allow adding/removing points and marking points as not visible.
  4. Extrinsic parameters (camera position and orientation) are computed once points are confirmed.
  5. Calibration results are saved as .toml and .qca.txt files; optionally, a .trc file for biomechanical use can be generated.

• Outputs:

  • Extrinsic camera parameters for 3D reconstruction
  • Calibration files compatible with Pose2Sim and OpenSim

Key Features

• Requires user interaction for accurate point matching.
• Depends on quality of calibration points and proper configuration.
• Intrinsic calibration must be provided externally (via Pose2Sim).
• Future improvements include automatic pattern detection and integrated intrinsic calibration.

1.5 Markerless motion analysis

Currently, this pipeline depends on Theia3D for the 3D pose estimation step. However, to reduce dependency and increase flexibility, it is desirable to develop an in-house markerless motion analysis solution.

For 2D pose estimation, Sport 2D can be utilized, which supports real-time 2D pose analysis(combined with GoPro Preview module). For 3D pose estimation, Pose2Sim is a promising tool. Pose2Sim accepts videos of varying lengths that do not require prior calibration or synchronization.

Since this pipeline generates synchronized videos with an identical number of frames and calibrated cameras, there is an opportunity to optimize Pose2Sim pipeline for this specific context. By leveraging the synchronized and calibrated videos, the Pose2Sim code can be restructured to perform pose estimation, person association, triangulation, and OpenSim simulation all within a single for-loop that iterates over the number of frames across all videos simultaneously.

This optimization would significantly accelerate the Pose2Sim workflow, enabling faster and more efficient markerless 3D pose analysis tailored to synchronized multi-camera data.

1.6 Report Generator

This module processes motion capture data(theia output) to extract biomechanical metrics and generate visual reports.

Process

The process loads motion capture data from a Theia .c3d file alongside corresponding segment information from a .json file.
It extracts key biomechanical parameters, including:

• Joint angles over time
• Segment lengths
• Distal marker positions

Outputs

• .trc – distal marker positions
• .csv – joint angles over time
• Multi-page .pdf – plots of joint angles