SimpleTrack-Analyzer (GoPro Telemetry Extraction and Overlay)

Overview SimpleTrack-Analyzer is a specialized Python tool designed to help novice track drivers achieve the most critical skill in performance driving: consistently maximizing tire friction. It extracts raw telemetry from GoPro and processes it to generate a real-time Friction Circle (G-G Diagram).

Technical Implementation & Logic Metadata Parsing (GPMF): The tool parses the GoPro Metadata Format (GPMF) stored in the MP4’s high-level payload. It extracts raw binary data for GPS (GPS5), Accelerometer (ACCL), and Gyroscope (GYRO) by traversing the FourCC-tagged telemetry track.

Data Synchronization Algorithm: Since sensors and video operate at different frequencies (e.g., IMU @ 200Hz, Video @ 60fps), I implemented a linear interpolation logic. This ensures that every video frame is mapped to the most accurate sensor state by calculating the precise temporal offset between the camera’s internal clock and the GPS timestamp.

Coordinate Transformation: Raw IMU data is processed to convert “camera-fixed” coordinates into “world-fixed” or “vehicle-fixed” metrics. This involves:

Noise Filtering: Applying a moving average filter to raw G-force data to remove high-frequency vibration noise (essential for FPV and automotive use).

Vector Projection: Calculating the resultant G-force and heading direction for the UI gauges.

Automated Rendering Pipeline: The system uses OpenCV to render dynamic gauges and FFmpeg for the final video compositing. It supports batch processing for long-duration recordings, focusing on maintaining 1:1 sync accuracy even in 4K/5K high-bitrate files.

Key Features Data Export: Supports converting binary GPMF to structured .csv or .gpx for further engineering analysis.

Visual Overlays: Real-time speed, G-force, altitude, and GPS track visualization.

Performance: Optimized for low memory footprint during the parsing of large video files.

View on GitHub:https://github.com/kkbin505/SimpleTrack-Analyzer

Tech Stack Python, OpenCV, FFmpeg, NumPy, Pandas.