Drag, drop, classify
Drop a JPG, PNG, GIF, or WEBP image and receive the model output without leaving the page.
CNN image classification for gemstone inspection
Gemstone authenticity, checked in seconds.
Upload a gemstone image and get a focused CNN prediction across emerald, ruby, turquoise, and imitation classes with confidence scores.
6
Supported classes
150
Pixel CNN input
Batch
Multi-image upload
Built for quick visual checks
A cleaner inspection workflow
Readable confidence
The result highlights the top class and keeps every probability visible for comparison.
Batch review
Process multiple images together and scan results in a compact grid built for review.
Model labels
Real and imitation classes are separated at a glance.
Use the exact labels returned by the API when validating or documenting model output.
Authentic gemstones
- EmeraldNatural green beryl
- RubyNatural red corundum
- TurquoiseNatural blue-green mineral
Imitation gemstones
- Fake_EmeraldGlass or synthetic imitation
- Fake_RubyRed glass or synthetic substitute
- Fake_TurquoiseDyed resin or polymer imitation
How it works
From image to prediction
The workflow is built for quick checking. Each step keeps the pipeline legible, from image intake to the final probability readout.
Fast, readable, and built for repeated review
01
Upload
Select one image or a batch from your device.
02
Preprocess
The API resizes and normalizes the image for the CNN.
03
Infer
The trained model scores all six gemstone classes.
04
Review
Inspect the predicted class, confidence, and probability spread.
Ready when the API is running