Home/AI Racing & Track/AI Marathon

AI Marathon
Personal race videos for every runner

A multi-vision recognition and automated video production pipeline that identifies each runner from large-scale race footage and generates personalized landscape videos, vertical clips, and media packs across the full race workflow.

Vision-only Matching

Combines face, body, and bib recognition without relying on a single timing point.

Runner Spotlight

Automatically highlights the target runner in crowded scenes for a stronger personal experience.

Automated Delivery

Generates landscape videos, vertical clips, and media packs for download, sharing, and editing.

40+
Event samples
300,000+
Runners served
92.2%
Generation rate
+101%
Download uplift with spotlight

Generation Workflow

An automated loop from race capture to runner delivery

01
Video Capture
Deploy cameras at start/finish areas, supply stations, and scenic route segments to capture multi-point race footage.
02
Live Stream Aggregation
Aggregate multiple video streams to the cloud for concurrent recognition, clipping, and task scheduling.
03
Multi-model Recognition
Combine bib recognition, face recognition, body detection, and motion trajectory to identify each runner across shots.
04
Personalized Clip Generation
Automatically edit templates with runner spotlight, race milestones, pace atmosphere, and celebration copy.
05
Runner Access
Reach runners through QR codes, messages, or event pages, with one-click download, sharing, and remixing.

Core Capabilities

Built around accurate recognition, stable rendering, and efficient runner access

Recognition at Any Checkpoint
Cameras can be placed at any high-value route segment, not only timing checkpoints, capturing more memorable moments.
Start/finish sprint
Supply station moments
Scenic route clips
Crowd atmosphere
Spotlight Tracking Effect
Track all runners first, then identify the target and apply smooth highlighting for stable visual focus.
Target highlight
Cross-shot smoothing
Re-identification
Shareability boost
Video Production Pipeline
Automates recognition, clipping, template rendering, transcoding, and distribution for mass-participation races.
Auto clipping
Template rendering
Landscape & vertical output
Media pack generation
Operations Console
Supports race setup, camera point management, job monitoring, retry handling, and data archiving.
Camera point setup
Job monitoring
Failure retry
Data archive
Performance Analytics
Track generation, download, sharing, and camera-point contribution to optimize future races.
Generation rate
Download analytics
Camera contribution
Sharing impact
Personalized Remixing
Runners can select clips, templates, and copy to create social-ready personal videos.
Clip selection
Template switching
AI copywriting
Social sharing

Vision-only matching for broader race scenarios

Traditional approaches often depend on timing checkpoints or external chip data, limiting footage to a few positions. Vision-only matching allows cameras at any valuable route segment and identifies runners through multimodal recognition, turning landmarks, supply moments, and finish-line scenes into personal video assets.

No single timing-point dependency
Supports non-chip race formats
Flexible multi-point deployment
Built for mass concurrent rendering
Anonymized race sample
Generation performance
Registered runners6,191
Personal videos generated5,580
Single-event generation rate90.13%
Download rate with spotlight effect51.3%

Use Cases

For road races, sports tourism, campus events, corporate wellness, and more

City Marathon
For full marathons, half marathons, and fun runs, creating personal race memories for every runner.
Campus & Corporate Runs
Low-barrier deployment for campus sports events, corporate wellness runs, and community races.
Sports Tourism
Combines landmarks and runner footage into shareable event content assets.
Vertical Marathon
Supports stair climbs, mountain routes, and special challenge courses with personal highlight clips.

Turn race footage into personal runner assets

For race organizers, operators, and venue owners, this product covers camera planning, video ingestion, AI recognition, automated rendering, and performance review.