Road-user detection
Identify cars, trucks, buses, bicycles, pedestrians, and other supported object classes in recorded traffic scenes.
Open source · AI traffic analytics
Turn traffic video into planning-ready data.
SimJam Computer Vision Analytics detects and tracks road users in recorded video, then converts their movement into counts, calibrated speeds, trajectories, and structured outputs for transportation planning and traffic-operations studies.
- YOLOobject detection
- CSVplanning-ready exports
- MITopen-source project
▶ Getting-started tutorial
The workflow
From roadway footage to usable evidence.
SimJam brings detection, tracking, transportation measures, and export into one workflow so planners can spend less time moving between disconnected tools.
-
01
Load traffic video
Begin with recorded footage from a CCTV, drone, roadside camera, or dashcam and define the scene to be studied.
-
02
Detect and track
Use a YOLO-based model to identify road users and maintain consistent object IDs as they move through the frame.
-
03
Analyze and export
Produce counts, trajectories, calibrated speed measures, summaries, and CSV outputs for further analysis or reporting.
What it measures
Computer vision designed around transportation questions.
Move beyond bounding boxes. SimJam organizes tracked road users into measures that can support traffic counts, operational evaluation, safety analysis, and simulation inputs.
Watch the short demoMulti-object tracking
Maintain consistent IDs over time so each road user's movement can contribute to counts and trajectories.
Speed and trajectory analytics
Estimate speeds when calibration is available and extract movement paths for operational or safety analysis.
Structured data export
Export CSV summaries and structured results for spreadsheets, reports, visualization, or simulation calibration.
Inside the application
Two connected stages. One analysis path.
Review detections and tracked movements first, then move into transportation analytics and export without rebuilding the analysis in a separate application.
Start here
Run your first traffic-video analysis.
The project documentation lists Python 3.12 or later and Visual Studio Code as requirements. Follow the video tutorial while keeping the repository open for the current files and licensing notes.
- 1Prepare the environment
Install a supported Python version and Visual Studio Code before opening the application files.
Review current requirements ↗ - 2Get the project
Download or clone the public repository and keep its application modules together.
Open the GitHub repository ↗ - 3Follow the guided workflow
Load sample footage, run detection and tracking, review the analytics, and export the results.
Watch the getting-started tutorial ↗
Licensing: the SimJam project is MIT licensed. Its Ultralytics YOLO dependency uses AGPL-3.0; review the repository before redistribution or deployment.
Planning applications
Use the outputs where transportation decisions are made.
Choose the measure that answers the study question, document the camera and calibration assumptions, and retain the exported data for transparent comparison.
Turning movements and approach volumes
Convert tracked movements into repeatable counts for intersection and corridor studies.
Speed estimation and distributions
Use calibrated footage to examine operating speeds and how they vary across the observation period.
Trajectory and near-miss analysis
Export road-user paths for deeper investigation of interactions and potential conflict patterns.
Before-and-after studies
Compare observed movement after traffic calming, signal timing, or another operational change.
Calibration and validation data
Prepare observed volumes, speeds, and movement patterns for microscopic traffic models.
CSV summaries and structured outputs
Move analysis results into spreadsheets, visualizations, technical reports, or data archives.
Common questions
Before you download.
Use the public GitHub issue tracker for reproducible bugs, feature requests, and questions about the current application.
Ask on GitHubWhat is SimJam Computer Vision Analytics?
It is an open-source traffic-video analytics application that combines road-user detection, multi-object tracking, transportation measures, and exportable results.
What types of video can I analyze?
The project is intended for recorded traffic footage such as CCTV, drone, roadside-camera, and dashcam video. Results depend on image quality, viewpoint, visibility, and the selected detection model.
Can it calculate speed?
Yes, when calibration is available to relate image coordinates and movement to real-world distance. Uncalibrated video should not be treated as a reliable real-world speed measurement.
What can I export?
The documented workflow supports planning-ready CSV summaries and structured results for measures such as counts, speeds, and trajectories.
Is SimJam Computer Vision free?
The SimJam source code is available under the MIT License. The application uses Ultralytics YOLO, which has its own AGPL-3.0 licensing requirements.
Ready to analyze traffic video?