✳ CropTrainLabW3DS AGRO
YOUR IMAGES. YOUR KNOWLEDGE. YOUR MODEL.

Any weed.
Better models.

Turn your field images and annotations into a focused training dataset. Start with the data in your eVault.

✳CROP & WEED
VISION LAB
Application foundation · eVault connection and training worker are not yet connected.
eID identityChecking…
Your eVaultsAwaiting connection
Training workerNot connected
01 / YOUR DATA

Build from your eVault

Choose a personal, company or group eVault you can access. Images and annotations stay with their owner.

▧

No dataset loaded

After connection, this view will show image counts, annotation classes, dimensions and checks needed before training.

No files are copied into a CropTrainLab database.
02 / TRAINING SETUP

Prepare your experiment

Use a DINOv2 foundation model with a task-specific training head. Choose the settings you need.

Preserve small seedlings at original detail. Split original captures into training and validation groups before making tiles.

Foundation weightsSelect compatible weights from your eVault after connection.

Downloading a plan does not start training or reserve GPU capacity.

Keep the detail

Tile full-resolution images for tiny seedlings, or scale entire images for broader patterns.

Check before training

Inspect labels, rare classes and capture groups before a model learns from them.

Choose your compute

The worker boundary allows local trials and future GPU experiments through the same interface.

Sign in with your eID

Scan with your eID Wallet.

eID sign-in QR codeOpen eID Wallet