STEP 1 OF 4Images
AI-powered
STEP 01
Tell us about your use-case
Describe your use-case and pick a domain — that's all you need. Optionally drop up to 4 reference frames and our AI will generalise classes, lighting, and variations directly from your images.
optional
Drop up to 4 sample images, or click to browsePNG, JPG, or WEBP · up to 8 MB each · more images = better AI generalisation
Domainoptional
STEP 02
What are you trying to detect?
We've seeded a starting set. Tap to keep or drop, and add your own. One class is enough to keep going.
0 selected
STEP 03
Dial in your variation needs
Pick the conditions and diversity dimensions that matter for your model. Then choose how big you want this batch to be.
Conditions
Weather, lighting, time of day.
Diversity
Subject types, environments, occlusion scenarios.
Camera angles & perspectives
Viewpoints and mounting positions to include in the dataset.
Volume
Roughly how big should this dataset be?
STEP 04
Your dataset card
A short, professional read on what you've configured. Good enough to share internally before we hop on a scoping call.
Synthetic Dataset Configuration
- Your custom synthetic dataset, scoped to the classes and variations you selected.
Detection classes
—
Conditions
—
Diversity
—
Camera angles
—
Volume
Standard (~1,000 – 2,000 images)