Detect bibs hidden under rain jackets, hydration packs, and darkness. Work from remote checkpoints with no connectivity. Process during events, deliver galleries next day.
Trail running and ultra marathons are the most extreme photography challenge: remote mountain locations with zero connectivity, runners spread over 100+ km of course, multiple checkpoint positions with 5-20 photos each, extreme weather (rain, mud, night sections with headlamp glare), and bibs hidden under technical gear. Photographers work from isolated positions, managing battery and storage with no access to power or data until returning to base camp.
Typical Event
12-48 hours (ultra marathons) or 6-12 hours (trail races)
Photo Volume
1,000-5,000 photos total across all checkpoints (50-500 per checkpoint)
Delivery
Next-day for social media, 3-5 days for full gallery (limited connectivity at finish location)
Key Challenge
Bibs hidden under rain jackets/hydration vests, no connectivity at checkpoints, night sections with headlamp glare, extreme weather damage to bibs and cameras
Download the starting list CSV from race organizers. Download RaceTagger's offline processing module to your laptop (available for download, no internet required). Test the offline setup on a sample batch to confirm it runs without internet. Plan your checkpoint positions with photographers (typically 2-4 checkpoints spread 25-30km apart). Each checkpoint team will collect photos and merge batches at base camp.
Pro tip
Remote checkpoints mean power is precious. Bring a portable solar charger and manage camera battery cycles carefully. Process photos in batches during the event (while runners pass at night and checkpoints are quiet) so you're not rushing processing at finish line at 2 AM.
Position yourself at 3-5 strategic checkpoints along the course. Shoot early (runners fresh, bibs visible) and late (night sections, headlamps) at each position. Don't expect to shoot the same runner twice unless it's a shorter loop course. Bibs are hidden: expect runners to wear rain jackets (even in dry conditions, for wind), hydration vests, and chest packs. Shoot chest height or slightly below for the flattest bib angle through the jacket. Anticipate 50-500 photos per checkpoint per event.
Pro tip
Trail runners prioritize gear over visibility. Most will have bibs hidden by mandatory safety vests or rain protection. Shoot multiple angles of each runner at each checkpoint. RaceTagger will pick the clearest frame. Accept that many bibs will be partially hidden — that's normal for trail running.
Create an event folder on your laptop running RaceTagger's offline module. Import photos from each checkpoint and the starting list CSV. Run offline batch processing — no internet required. Processing runs locally at ~4 seconds per photo, same speed as cloud. For 3,000 checkpoint photos spread across a 24-hour event, process in 2-3 hour batches during quiet periods (overnight while runners sleep at aid stations).
Pro tip
Don't wait until the finish to process. Batch during the event when you have downtime at checkpoints. By the time the last runner finishes, your photos are already tagged and reviewed. You'll be ready to deliver next morning, not tomorrow night.
RaceTagger flags low-confidence detections. In trail running, expect 20-30% flagging (significantly higher than road events) due to jacket occlusion and weather damage. Review only the flagged shots. For completely hidden bibs, don't force a guess — mark as 'can't determine'. Use checkpoint position and timing to assist: if you know Runner A passed Checkpoint 2 at 10:47 AM and Runner B passed at 10:50 AM, context helps resolve ambiguous middle photos.
Pro tip
Trail runners understand and accept that some bibs won't be readable due to weather and gear. Deliver 70-75% accurately tagged photos next day. That's a WIN. Don't over-engineer the 25-30% untaggable shots — focus on delivering good-quality galleries quickly.
Once you have internet access (base camp or hotel), sync your offline-processed photos to RaceTagger cloud. Export tagged metadata as XMP sidecar files. Merge checkpoint photo galleries by runner (same runner may appear at checkpoints 1, 3, and 5). Create a 'runner' folder for each participant with all their checkpoint photos chronologically ordered.
Pro tip
Many ultra marathons have a finish line gallery and checkpoint photo packages sold separately. Organize by runner first, then by checkpoint. This lets you offer 'Checkpoint Package #2' (all runners at CP2) or 'Personal Album' (all checkpoints of runner #247).
Publish next-day gallery with checkpoint-by-checkpoint organization (runners see all available photos of themselves across the race). Offer checkpoint photo packages (all runners at CP2) and personal runner albums (all checkpoints of one runner). Ultra runners are accustomed to delayed delivery — next day is professional, next week is acceptable. But deliver ASAP and you'll earn repeat contracts.
Pro tip
Ultra marathons build tight communities. Do a GOOD job (complete galleries, accurate tags, nice presentation) and those runners will hire you year after year. Speed is nice-to-have, quality is essential.
Why it's hard: Trail runners layer heavily or wear mandatory event safety vests. Jackets cover the entire torso. Hydration backpacks cover bibs from the shoulders down. Only the top 2-3 digits visible, or nothing at all. The bib is physically occluded.
How AI helps: RaceTagger is trained on occluded trail bibs and uses position inference to estimate full numbers from partial visibility. Confidence is lower (75-82% vs 95% on road races). Flagged photos are reviewed with context (timing, position in field). Many shots simply can't be read — that's okay for this sport.
Why it's hard: Paper bibs aren't designed for all-day mountain rain. Ink bleeds, paper tears and curls, numbers fade to gray. By mile 50 of a 100-mile ultra, many bibs are illegible. Weather damage compounds at night.
How AI helps: AI handles faded and damaged bibs better than OCR (uses context, not just character matching). But there's a physical limit: if the ink is completely gone, no AI can read it. RaceTagger flags these honestly as 'unreadable due to damage.'
Why it's hard: Night running means runners have headlamps on, creating lens flare and harsh shadows. The bib is brightly lit or completely dark depending on lamp angle. Contrast between number and background disappears. Extreme low-light photography.
How AI helps: AI is better at low-light bib detection than traditional OCR (understands context even in darkness). Still, night bib reads are flagged for review (confidence 70-80% instead of 95%). Accept night photos as 'best effort' rather than 'guaranteed accurate.'
Why it's hard: On 40% grades and technical terrain, runners are bent at the waist (climbing), or sitting on rocks (resting). The bib is foreshortened, wrinkled by torso angle, sometimes folded at checkpoints where runners rest. The flat paper becomes curved and illegible.
How AI helps: AI understands body posture and can read foreshortened bibs from steep-angle photos. This is where AI vision beats OCR clearly — context matters. But extreme folding still causes low confidence.
Why it's hard: Mountain races often have checkpoints with zero cell signal, no WiFi. Photographers can't upload to cloud for processing. Traditional cloud-based tools are useless. You need offline capability.
How AI helps: RaceTagger offers offline mode: download the processing engine, run it locally on your laptop, process without internet. Sync to cloud when you reach a connected location (base camp, hotel). Solves the connectivity problem completely.
Manual Tagging
6-10 hours for 3,000 checkpoint photos (photography team + post-event taggers working at base camp)
70-80% on hidden bibs, 85-90% on visible ones. Fatigue drops accuracy further after few hours.
With RaceTagger AI
~3 hours for 3,000 photos (batch processing during event + next-day export)
75-82% on hidden bibs (flagged for review), 90-95% on visible. Accept 20-25% flagging as normal for this sport.
Real-world scenario
You're covering the UTMB (Ultra Trail Mont-Blanc), a 170km mountain race with 10,000m elevation gain. You position photographers at 4 checkpoints: Chamonix (KM 0, start), Les Contamines (KM 34), Les Chapieux (KM 66), and Rundet (KM 100). Each checkpoint team shoots 200-400 photos of passing runners. Over 40 hours, you accumulate 1,200 photos across checkpoints. By midnight on Day 2 (while the final runners are still climbing), you've batch-processed all photos offline on your laptop (3 hours of processing spread over 40 hours). You've reviewed flagged photos at each checkpoint in downtime. By base camp checkout (4 PM on Day 2), you have 900 of 1,200 photos accurately tagged (75% accuracy, normal for ultras). Untaggable photos (completely hidden bibs, damaged from 48+ hours of rain) are marked 'no bib visible' — runners understand. By next morning, galleries are live. Runners who finished 36 hours ago are already sharing photos on social media, telling their friends. Your galleries generate sales momentum because they're FRESH, not cold.
Next-day delivery on a 100-miler is industry-leading. You become the preferred photographer for UTMB and similar ultras. Offline processing means you didn't need hotel internet or base camp power — you worked autonomously. That efficiency wins the contract year after year.
500 free tokens. Download offline processing, upload checkpoint photos from your last trail event, and process without internet — see the results.
Start tagging for free →How does RaceTagger process photos when there's NO internet at the checkpoint?
RaceTagger has offline mode: download the processing engine before the event, run it locally on your laptop without any internet. Process photos at checkpoints during quiet times (runners sleeping at aid stations). Sync to cloud and export when you reach internet access at base camp or hotel.
What accuracy should we expect for trail running bibs hidden under jackets and vests?
75-82% on jacket-hidden bibs, with honest flagging for ambiguous reads. This is significantly harder than road racing (95-97%). Accept flagging as normal — you'll review flagged photos manually. 25-30% of total photos will be flagged or untaggable due to occlusion. That's the nature of trail running gear.
Can we use RaceTagger for ultra marathons that span 24+ hours with runners passing checkpoints at night?
Yes, but budget extra flagging at night. Headlamp glare and low light reduce AI confidence to 70-80%. Daytime checkpoint photos are clearer (88-93%). Overall expect ~20-30% flagging across a 24-hour event, with heavy flagging on night sections.
What if a runner's bib is completely destroyed (soaked, illegible, torn off) and we can't read it at all?
Mark it as 'can't determine' or 'bib not visible'. Don't guess. Trail runners understand that extreme weather can damage bibs beyond recovery. Deliver honest galleries: 'We got a great shot of you at CP2, but the bib wasn't readable due to weather.' Runners appreciate the photos regardless.
Can we offer separate packages for each checkpoint, or should galleries be organized by runner?
Both. Organize your delivery as: (1) Personal Runner Albums (all checkpoints of one runner, chronological), and (2) Checkpoint Galleries (all runners at one checkpoint). This lets runners buy 'their journey' or 'the entire CP2 experience with friends.'
Related
Offline Photo Processing for Remote Trail Races — No Connectivity NeededTechnical deep dive on RaceTagger's offline mode, deployment, and workflow for photographers working at remote checkpoints with zero internet
Related
Night Section Photography: Headlamp Glare and Low-Light Bib DetectionSpecific guidance on shooting and tagging during night sections with headlamp glare, expected accuracy rates, and manual review strategies
Related
How Ultra Marathons Scale Your Photography Business — Multi-Checkpoint WorkflowBusiness-focused guide on how multiple checkpoint positions, next-day delivery, and checkpoint packages generate higher revenue per event