Is your crop photo AI-ready?
Upload a field photo and get an instant verdict. We first confirm there's a crop and no weeds, then check the photo is sharp and genuine, and only then screen it for disease and pests.
Drop photos or a whole folder here
JPG, PNG or WEBP · up to 15 MB each · up to 500 photos per batch
Tip: you can also paste an image with Ctrl + V
Three quick gates, one clear answer
Each step only runs if the one before it passes, so you get a fast answer and never a health result you can't trust.
Is it a clean crop photo?
First we confirm the photo actually shows a crop, not soil, sky or people, and that it isn't a photo of weeds.
Is it sharp and genuine?
Next we check the photo is in focus and was captured in the field, not photographed off a screen.
Is the crop healthy?
Only good photos reach this step, where we screen for disease and visible pests.
Six specialised AI models
Each check is a dedicated deep-learning model, built for real-world field photos.
Get it right the first time
Four habits that get almost every field photo through all three gates.
Fill the frame
Get close so the crop or affected leaf takes up most of the photo.
Tap to focus
Tap the leaf on your screen and hold still for a second before shooting.
Use daylight
Shoot in natural light. Avoid harsh backlight and deep shade.
Shoot the real plant
Never photograph a screen, print-out or another phone's display.
Ready to check your photos?
It's free, takes seconds, and works for a single photo or a whole folder.
Use the same models in your own software
Everything this page does is available over HTTPS. Send a photo, get JSON back. Run the full audit, or call any one of the six models on its own.
- Base URL
https://audit.annam.ai- Auth
X-API-Keyheader on every request- Input
- multipart
fileor JSON{"image": base64}· JPG, PNG, WEBP up to 15 MB - Limits
- 60 requests / min per IP · 20 images per batch
When something goes wrong
400No image, or the file isn't a readable image.401Missing, wrong or revokedX-API-Key.413Image over 15 MB, or more than 20 files in one batch.415Not a JPG, PNG or WEBP file.429Rate limit reached. Wait a few seconds and retry.503That model is temporarily unavailable.
Getting a key
- Email us with your organisation, what you're building and roughly how many images a day you expect.
- We issue a key for your project.
- Keep it on your server. Never put it in a mobile app or web page.
-
Vedant AgnihotriMachine Learning Engineerstaff.vedant.agnihotri@iitrpr.ac.in
-
Deepak MeenaLead Software Engineerstaff.deepak.meena@iitrpr.ac.in
Questions, answered
Who is Data Audit for?
Field teams, researchers and anyone building crop image datasets. It helps you reject unusable photos at the source, so your AI models are trained on clean, genuine field data.
Why were some checks skipped?
Checks run in order. If no crop is found or weeds are detected, nothing else is checked. If the photo is blurry or a screen capture, the health checks are skipped, because an unclear photo can't give a reliable health result. Retake the photo using the tips shown and try again.
What does the confidence percentage mean?
It's how sure the model is about its answer. Above 90% is a strong signal. Results between 50% and 70% are borderline, so a second photo from a different angle is a good idea.
Can I use this to diagnose my crop?
Treat the health results as an early screening, not a diagnosis. AI can make mistakes. For treatment decisions, please consult a local agronomist or Krishi Vigyan Kendra.
Can I audit a whole folder?
Yes. Click Upload a folder (or drag a folder onto the page) to audit up to 500 photos in one go, including sub-folders. Photos are checked one after another, and you can pause at any time. When it's done, download a CSV report listing every file name with its verdict and the reason.