Dive-centre photo transects, scored by AI, that show if a reef is recovering
TL;DR Reeftally trains dive centres in the Andamans and Lakshadweep to shoot the same photo transects every quarter. Software scores live coral, bleaching and algae from the images. Coral restoration funders and marine park managers pay for dated reports that show whether the reef is getting better or worse.
India has roughly 2,330 km² of mapped reef across four regions, most of it around the Andaman and Nicobar Islands and Lakshadweep[2]. Heat stress is getting worse. In 2024 Lakshadweep logged 9.2 degree heating weeks, the highest ever recorded there, and even blue coral bleached at 60%[1].
The weak link is data. A FAO review notes that India lacks reliable baselines for live coral cover, and that Andaman monitoring has covered only a few sites[2]. Surveys need boats, fuel, gear and trained divers, so donors rarely pay for them year after year[3].
Restoration projects do get funded. But how many can show a funder what happened two years later? Very few.
What it does, in regulatory terms: non-extractive observation only. Nobody touches or collects coral. Corals are protected wildlife in India, and diving in marine national parks needs forest-department permission, so every survey site is agreed with the park first.
Reeftally fixes permanent transects at each site and trains staff at partner dive centres to photograph them on a set schedule. Paying guests can join as a supervised "survey dive", which covers part of the boat cost.
Images upload after the dive. A model trained on Indian reef photos estimates hard coral, bleached coral, algae and rubble, and a marine biologist checks a sample of every batch. The buyer gets a dashboard and a quarterly report per site.
The seed wanted autonomous underwater swarms and I dropped them. Robots cost more, need more permits and still miss what matters most here - cheap, repeated pictures of the same spot. Divers are already in the water every day anyway.
Two buyers carry it. Coral restoration projects funded through CSR need proof that planted fragments survive, and today that proof is often a handful of photos. Marine park managers need trend data across many sites but have small survey budgets.
Research groups buy data access, and dive centres earn a fee per survey and sell the survey dive to guests. I found no public price for reef surveys in India, so pricing has to come from interviews.
The 2024 bleaching was the most severe on record in Lakshadweep[1], which puts reef condition on funders' agendas. Image-based scoring is routine now - one free academic annotation platform hosts more than 5.8 million reef images and 304 million point annotations[4]. A diver-led protocol for the Andamans was already drafted in 2016, so the method has local roots[3].
There's no public figure for what India spends on reef monitoring. The narrow signal is that restoration and monitoring work already runs in the Andamans through long-running marine conservation trusts[5], and each of those projects needs before-and-after evidence.
The adjacent number is the reef area itself, about 2,330 km² across four regions[2], which sets a ceiling on how many sites you can sell.
A marine conservation trust runs restoration and monitoring in the Andamans[5]. Free academic platforms offer image annotation, which is a tool and not a survey service[4]. Government institutes run their own surveys at selected sites.
The gap is a paid, repeatable service that turns dive-centre visits into comparable data a funder can actually read.
| Difficulty | Moderate (43/100) |
| Starting capital | ₹40–75 L |
| Time to launch | 6–9 months |
The software is the easy part. The hard work is keeping photo quality steady across dozens of divers, getting park permission site by site, and training a model that tells bleached coral from sand in murky water. Selling data to buyers who are used to free academic surveys takes a lot of patience too.
There is no licence to run reef surveys as a business, but where you dive is tightly controlled.
Building this outside India? The categories above usually have a local equivalent, so check with your own regulator.
Don't trust the model on its own for bleaching calls or recovery claims. Murky water fools it, and funders will make decisions on these numbers, so a scientist has to check every batch.
Want to work out the AI layer for Reeftally? Talk to Yash →
Here's what can go south -
I'd back a marine biologist who dives and can sell to CSR teams. If you're hoping for a robotics business then this isn't it.
First prove two things - one restoration funder will pay for quarterly reports across three sites, and dive-centre images score within a few percentage points of a scientist's survey.
Plan for about ₹40–75 L, mostly for cameras, training dive-centre staff, model building and a marine biologist's time.
Yes, inside marine national parks you need forest-department permission, and corals are protected wildlife. Agree every site with the park before you start.
Yes, the dive-centre model works anywhere with active reef diving. Check that country's marine park permits and wildlife rules first.
The free Reeftally validation kit has the line-by-line capital breakdown, the effort scoring, the permits and data checklist, a 14-day validation sprint with pass/fail targets, a 90-day launch plan, 10 interview questions for funders and dive centres, and 5 assumptions to test. Download the kit →
Thinking about building Reeftally or something close to it? The make-or-break step is finding a funder who will pay for evidence, not only for planting, and I'm happy to help you work out who that is. If you need a hand, you know where to find me. Talk to Yash →
For illustration only. Names are fictional and ideas hypothetical; sourced facts are real but may be out of date. Nothing here is legal, financial or professional advice, and Yash Khatri is not personally liable for any use of it. Read the full disclaimer.