Survey an Orchard With Just a Phone

You don't need a drone or a survey crew to digitize an orchard. A modern phone tags every photo with GPS coordinates, and that's enough to reconstruct where every tree is and what's on it. The method below is the one that produced a full durian-orchard twin from a single afternoon of walking.

Turn on location tags before you start

This is the step everything else depends on. In your camera settings, allow the camera app to access location, so each photo's EXIF metadata carries latitude and longitude. Take one test shot and check that it records a location. A survey of untagged photos can't be placed on a map — and you usually discover that after the walk, when it's too late.

Walk a pattern, not a wander

• Go row by row in one direction so you don't miss or double-cover trees.
• Stop at each tree and take a few frames from different angles — trunk, canopy, and any fruit or flowers.
• For tree crops where fruit and flowers sit low or on the bark (durian, jackfruit, cacao), shoot the trunk deliberately, not just the leaves.

A few photos per tree beats one perfect photo: extra angles give a counting model more to work with, and overlap is easy to deduplicate later.

Mind GPS accuracy under the canopy

Phone GPS is accurate to a few metres in the open and drifts more under dense canopy. That's usually fine — trees are metres apart — but it's why clustering photos into trees takes a little judgment rather than blind coordinate-matching. Standing at the trunk when you shoot, rather than 10 metres away zooming in, keeps the location honest.

From photos to trees

Back from the field, the photos cluster by location: frames taken within a few metres of the same trunk belong to the same tree. That collapses a few hundred photos into a few dozen distinct trees, each with its own coordinates and photo set. From there a vision model can count fruit and flowers per tree, and you have a queryable inventory instead of a camera roll.

Repeat it

The real payoff is the second survey. Because the method is repeatable, walking the same rows next season produces comparable numbers — the same trees, counted the same way — and now you can see change: which trees flowered, which lagged, how the block trended. A one-off survey is a snapshot; a repeatable one is a baseline.