To identify a plant problem from a photo, take three pictures, not one: the whole plant where it lives, a close-up of the worst leaf, and the underside of that leaf with the stem base in frame. Then read the damage by its pattern, and check the compost by hand, because no photo can show moisture or roots.
A photo shows scale insects, mildew and slug damage clearly. It cannot tell you that the compost has been wet for a fortnight, that the roots have gone soft, or that the plant moved closer to the window last week. Most wrong diagnoses, from people and apps alike, happen because the deciding clue was outside the frame.
What photos should you take to diagnose a plant?
Take three or four photos, and give each one a job. A lone close-up of a brown patch is the least useful picture you can take, because it hides position, pattern and conditions.
- The whole plant, in its actual spot. This shows which leaves are affected, which side faces the light, and what is nearby: a radiator, a draughty door, a saucer full of water.
- A close-up of the worst leaf, top side. Fill the frame and tap to focus on the damage. Blur turns fine speckling into vague discolouration, which is how pests get mistaken for nutrient problems.
- The underside of the same leaf. Spider mites, thrips, scale insects and mealybugs mostly live underneath. Almost nobody takes this photo, and it often settles the question.
- The base of the stem and the compost surface. A soft, dark stem base, white fuzz, small flies, or compost shrunk away from the pot wall are all visible here.
If you can, add a healthy leaf from the same plant, as a baseline for normal colour in that light.
Then make each photo count:
- Daylight, no flash. Flash flattens texture and hides mottling. Bright indirect light is ideal; harsh sun can make a healthy leaf look scorched.
- One plant per photo. Several plants in one frame make identification much harder, for you and for software.
- Zoom rather than lean in. Phones struggle to focus a few centimetres from a leaf; step back and use 2x.
What can a photo show, and what can't it?
A photo is excellent at surface patterns and close to useless for anything you would find by touch, weight or memory. Knowing which side of that line a problem sits on tells you how far to trust a photo-based answer.
| A photo can show | A photo cannot show |
|---|---|
| Where the colour change sits: old leaves, new leaves, between the veins | How wet the compost is a few centimetres down |
| Spots, rings, halos, powdery or fuzzy growth | Whether the roots are firm and pale or brown and soft |
| Pests big enough to see: scale, mealybugs, aphids, webbing | Spider mites and thrips, unless the shot is very sharp |
| Holes, notches and slime trails | How heavy the pot is |
| Wilting, stretched growth, leaf drop | What changed in the last fortnight: a move, a repot, the heating going on |
| Which side of the plant is affected | Night temperatures, draughts, or a radiator under the shelf |
Watering problems in particular are decided below the compost line, which is why telling overwatering from underwatering starts with a finger in the pot, not a look at the leaves.
What do common plant problems look like in a photo?
Most problems leave a signature: a position, a shape and a texture. This table covers the ones you can genuinely spot in a picture, and the check that confirms each.
| What the photo shows | Likely cause | Confirm it by |
|---|---|---|
| Soft, limp yellowing on old and new leaves at once | Overwatering or waterlogging | Compost wet a few cm down; heavy pot |
| Dry, crisp brown edges; lowest leaves yellowing first | Underwatering | Light pot; compost dry and pulling from the pot wall |
| Wilting with a dark, soft stem base | Root or stem-base rot | Tip it out: brown, mushy roots that pull apart |
| Fine pale speckling on the upper surface, sometimes fine webbing | Spider mites | Check the underside with a hand lens |
| Pale or silvery mottling with tiny black dots | Thrips | Undersides and flowers; the dots are their droppings |
| Small brown, white or yellow bumps on stems and veins; sticky leaves below | Scale insects | The bumps lift off with a fingernail |
| White, fluffy wax in leaf joints and crevices | Mealybugs | Look where leaves meet the stem |
| Clusters of small insects on soft shoot tips; sticky or sooty leaves | Aphids | Check new growth and the undersides of young leaves |
| Neat notches bitten from the leaf edges | Adult vine weevil | Creamy, C-shaped grubs in the compost. The RHS notes the root-eating grubs, not the notches, are the real damage |
| Irregular holes, ragged edges, silvery trails | Slugs or snails | A torch check on a damp evening |
| White, powdery coating on the leaf surface | Powdery mildew | It sits on the surface rather than in the leaf |
| Fuzzy grey mould on soft growth or old flowers | Grey mould (botrytis) | Usually damp, still air and dead material nearby |
| Bleached or pale brown patches on the sun-facing side only | Sun scorch | Was it moved somewhere brighter recently? |
| Dark, water-soaked leaves turning black after a cold night | Cold or frost damage | Check the overnight low for your area |
| Raised corky blisters, mostly underneath | Oedema | Wet compost plus cool, dim conditions |
| Yellow between green veins | Magnesium (old leaves) or iron (new leaves) | Which end of the plant it started at |
How do you read the damage pattern?
Ask three questions: where on the plant, what shape, and how fast. Together they do far more work than the colour alone.
- Where? Oldest, lowest leaves first usually means the plant is rationing: nitrogen, magnesium, overwatering or plain age. Newest leaves first points to something it cannot move around, such as iron. One side only suggests light, cold or a draught. Everything at once suggests the roots or a sudden change. The yellow-leaves walkthrough covers this in detail.
- What shape? Notches from the edge, holes in the middle, fine stippling, spots with a defined edge and patches between the veins are all different problems. Shape is often the fastest way to separate a pest from a disease from a growing-conditions issue.
- How fast? Overnight change points to cold, heat, a sudden move or a plant that ran dry. Change over weeks points to roots, nutrients, light or a pest population building. Compare with an older photo if you have one: a picture from last week is a fixed reference in a way memory is not, which is why slow problems are so easy to miss.
Why do plant apps and AI sometimes get it wrong?
Usually because the deciding clue was never in the photo. That limit applies to an expert looking at your picture too.
- Look-alikes. Overwatering and underwatering both cause drooping and yellowing. Mite stippling can read as a nutrient problem, and scale can look like part of the stem.
- Missing context. When it was last watered, or whether the heating just came on, is not in the pixels.
- Poor input. Blur, flash, several plants in one frame, or only the top of a leaf when the evidence is underneath.
- Confidence is not accuracy. A specific disease name sounds authoritative. Treat any name, from any source, as a hypothesis until the pattern and a hands-on check agree with it.
How should you check a diagnosis before acting on it?
Match it against the pattern, confirm it by hand, then start with the cheapest fix you can undo.
- Does the pattern fit? Position, shape and speed should all agree, not just the colour.
- Do the hand checks. Compost a few centimetres down, the weight of the pot, the underside of a leaf, the base of the stem.
- Start with reversible fixes. Stop watering, empty the saucer, move it out of direct sun, or separate a plant with pests. Do not reach for a spray until you have seen the pest or the growth it names.
- Change one thing, then re-photograph. Same angle, a week later. Judge by new growth, because damaged leaves rarely recover their colour.
- Ask a person if two rounds have not helped. An experienced gardener, a good independent nursery, or the RHS's gardening advice service if you are a member.
How does Mali diagnose a plant from a photo?
Mali is built around the same limits, so it tries to say when a photo is not enough rather than guess.
- Identifying a plant: you can add up to four photos of the same plant, since more angles give a better match. You see the top match with a percentage and the closest alternatives, and a weak match is labelled as a best guess.
- Health check: add one to a scan and Sidney, Mali's AI plant expert, names the disease or pest when she is confident, with a plan split into prevention, a natural remedy and a stronger treatment. When she is not confident, she says she cannot be sure from that photo and asks for a close-up check-in instead.
- Photo check-ins: the photo is read together with the plant's record (when it was last watered and fed, how far along it is, notes from recent check-ins) and your local weather. That context is what lets an answer take into account that you watered three days ago, rather than guessing from the leaf alone. You can then keep asking Sidney questions about it.
It cannot feel the compost or see the roots, so you still do those checks. Mali is free to use, the AI features come with a free allowance, and an optional Mali+ subscription lifts the limits. It is on iOS and Android, and there is more about how it works on the Mali home page.
Sidney, our resident purple snail, would like it on the record that if the leaf has ragged holes and a silvery trail, she has an alibi. She also suspects the slugs.
FAQ
Can AI identify plant diseases from a photo?
Often, for problems with a clear visible signature, such as powdery mildew, scale insects, mealybugs, sun scorch or slug damage. It is less reliable for watering, root and nutrient problems, where the evidence is under the compost or in the plant's recent history. Use it to narrow things down, then confirm by hand.
What is the single best photo for diagnosing a plant?
If you can only send one, make it a sharp close-up of the worst leaf in daylight, with a healthy leaf beside it for comparison. If you can send two, add the whole plant in its usual spot, so the pattern and the conditions are visible.
Can a photo tell if my plant is overwatered?
Not reliably on its own. Overwatering and underwatering produce very similar drooping and yellowing, and the difference is in the compost and the weight of the pot. A photo can raise the suspicion; your finger a few centimetres into the compost settles it.
Should I spray my plant based on an app's diagnosis?
Not on a guess. Confirm the pest or disease yourself first, and try the reversible fixes before any chemical treatment. The wrong spray costs time while the real problem carries on.
Do I need a macro lens to photograph plant pests?
No. A steady phone at 2x zoom in good light is enough for most pests. For spider mites, the RHS recommends a x10 hand lens, which is cheap and far better than any phone camera at that size.
Keep reading: how often to water houseplants in winter, how to protect plants from frost, or browse all posts.
— Akhil Madan, founder of Keeda Studios. I take far too many photos of my plants, which is why Mali starts with one.