Agronomist visual census
Human eyes in the field
The gap
- Symptoms only visible at 60–70% tissue infection
- Subjective — results vary inspector to inspector
- 5 agronomists cannot cover 5.7 M palms
Smarter targeting of which palms to inspect.
AI-based visual screening for Ganoderma (basal stem rot) — an offline Android app and management dashboard for plantation workers, agronomists and management.
- Green Routine monitoring
- Yellow Verify
- Red Inspect first
GanoAI
offline screening
Centre & frond detection
Shorter / irregular fronds flagged
Visual analysis result
Verification required
AI that shows your team which palms to inspect first
GanoAI is an agritech solution developed by Tkraf that enables fast and efficient Ganoderma screening through a smartphone app. Its AI model is trained with expert input to support screening accuracy and help plantations identify palms that require further PCR testing in a laboratory. This targeted approach helps save time, optimize testing costs, and support informed decisions on palm health.
Vision
To shape a future where technology empowers every oil palm plantation to stay ahead of Ganoderma, safeguarding productivity and long-term sustainability.
Mission
To equip oil palm plantations with fast, objective, and affordable Ganoderma scanning technology, enabling informed decisions and timely, targeted action to protect palm health.
-
Offline Android app
Field officers photograph each palm; analysis runs on the phone, no signal needed.
-
Crown-geometry AI
Computer vision measures frond direction, spacing and crown symmetry.
-
Order triage
Green / yellow / red priority, with photo and GPS for every palm.
-
Management dashboard
Maps, review queues and history from block to region level.
Ganoderma is a silent killer of oil palm
- USD 500Mestimated yearly losses to basal stem rot for producers in Southeast Asia
- 60–70%of tissue already infected when the first visible symptoms appear
- 3–5 yrslife expectancy of a mature palm once infected — a total loss of yield
- 860,610 haof mature oil palm projected to be wiped out in Malaysia by 2040
Sources: Remote Sensing (MDPI) 2022, 14(5):1239 · Frontiers in Plant Science 2023, 10.3389/fpls.2023.1156869
Millions of palms. Too little expert time
Hypothetical example for a large plantation — to be tailored with partner data.
40,000hectares of plantation
→
5.7 Moil palms (≈143 per ha)
÷
5agronomists available
=
1.14 Mpalms per agronomist
≈ 4,560
palms per agronomist, every working day, just to look at each palm once a year (250 working days).
The obstacle is not agronomic expertise.
The number of palms far exceeds the daily capacity for in-depth inspection — so census becomes rushed, sampling becomes random, and early cases slip through.
Today's detection methods each leave a costly gap
PCR lab test
Molecular confirmation
The gap
- Expensive, laborious stem sampling + lab work
- Days to results; not usable in the field
- Impossible to run on millions of palms
Satellite / remote sensing
Imagery from above
The gap
- Coarse resolution, cloud cover, long waits
- Weak at early stage — 57% test accuracy in a UAV study
- Confuses disease with nutrient deficiency
Sources: Frontiers in Plant Science 2023; Remote Sensing (MDPI) 2022, 14(5):1239.
Screen every palm first. Experts make the call
AI results set inspection priority — they do not declare a definitive disease status. Every palm gets a colour, a photo and a GPS point.
-
Green
Low indication
Stays under routine monitoring. Does not mean proven infection-free.
-
Yellow
Verification required
An anomaly, or the AI is not yet certain. Enters the review queue.
-
Red
High priority
Strong visual indication that needs prompt review by an expert.
Measuring crown patterns, not guessing a diagnosis
Computer vision converts visual characteristics into traceable, repeatable screening signals.
-
01
Frond geometry
Direction, distance and angular distribution of every frond from the crown centre.
-
02
Crown asymmetry
Detects imbalance in growth direction — a visible stress pattern of basal stem rot.
-
03
Patterns of change
Compares observations of the same palm across census rounds.
-
04
Triage priority
Converts signals into an inspection order for agronomists.
From one photo to a documented decision
Designed to slot into the existing plantation census — no new team, no new hardware beyond an Android phone.
- 01
Photo officer
Captures the crown from 4 sides following the initial SOP.
- 02
AI screening
Crown pattern & geometry processed on the phone itself.
- 03
Priority
Green / yellow / red, stored with photo and GPS.
- 04
Agronomist
Field inspection, follow-up and documentation.
Offline by design Scan deep in the estate with no signal — data syncs automatically when a connection returns.
Same budget: 143,000 palms screened vs 84 PCR tests
PCR lab test
Rp 708,000
per palm sample — DNA isolation + PCR, lab fee only
≈1,700×cheaper per palm
GanoAI screening
≈ Rp 600
per palm per census round (Rp 80,000/ha ÷ 143 palms)
- PCR on every palm
- Rp 318.6 M
- GanoAI screen + PCR on flagged 10%
- Rp 32.0 M
≈ 90% lower lab spend per hotspot
PCR price: Universitas Brawijaya Rector Regulation No. 66/2024, industry rate for DNA isolation + PCR; excludes stem sampling, labour, primers and transport. GanoAI price excludes VAT, onboarding and devices. 10% flag rate is illustrative. GanoAI targets PCR — it does not replace it.
Put expert hours and PCR budget where it counts
10×
fewer palms for agronomists to inspect in depth — more time and better judgment on each high-risk palm.
PCR savings formula
Samples safely reduced × actual cost per sample
GanoAI does not replace PCR — it targets it. Reductions are applied only after missed-case risk is assessed.
Layered accuracy: AI reach + expert eyes + PCR
- 1
GanoAI screens every palm
Objective crown-geometry measurement. The same photo gives the same score — no fatigue, no inspector-to-inspector variation.
- 2
Agronomists verify yellow & red
Expert eyes on far fewer, higher-risk palms: more time per palm means better field judgment.
- 3
PCR confirms targeted samples
Lab certainty where it changes decisions — including green control samples.
What your company gets
- Coverage of every palm, not a random sample
- PCR-grade certainty where decisions are made
- Fewer missed cases from rushed census
- Auditable record: photo, GPS, score, review
Each layer does what it is best at — the combined system is more accurate and far cheaper than any single method used alone.
Field trial experiment: documentation
Ganoderma palms
Healthy palms
Every palm tagged
Palm ID, sampling date & time and expert status (sakit / tidak sakit) recorded on site.
Blind comparison
Expert assessment is recorded independently, then compared with GanoAI screening.
Positives and controls
Symptomatic palms are tested alongside healthy controls in the same estate.
Tested with experts. Measured with agreed KPIs
Preliminary blind test
GanoAI vs agronomist assessment on a set of palms.
Follow-up evaluation
Consistency of results and field workflow readiness.
Cross-plantation validation
Independent controls + PCR samples + different plantations.
Sensitivity
Share of reference-positive palms flagged yellow or red.
Productivity
Analysable palms per staff-hour, incl. retakes and walking.
Missed cases
False negatives estimated in independently checked green samples.
Realistic cost
Cost per screened palm, equipment readiness, value scenarios.
Let's put it to the test on your plantation.
3 agreements to start the pilot
- Select 2–3 blocks and a field supervisor
- Agree on the expert method, PCR protocol and KPIs
- Finalise pilot schedule, logistics and costs