Object detection software. No ML team required.
Train object detection, classification, segmentation or anomaly models from 10 labeled images. Test on new images and video, then call it over the API.
Object detection models and more, for every inspection question.

Classification
“Is it OK, cracked or reject?”
10 labeled images · 2+ classes
Object detection
“What's wrong, and where?”
10 labeled images · 1+ class
Segmentation
“Exactly which pixels?”
10 labeled images · 1+ class
Anomaly detection
“Does it look different from good?”
Good parts only · UltraTrain computer vision models: three choices, then Start Training.
Pick a type, a time budget (Fast, Normal or Extensive) and a size (Mini on CPU or Plus for production).
- Saves a version first
- Runs in the background
- Plus on Pro and Ultra; Extensive on Ultra
Test your computer vision model on images it has never seen.
Run it on up to 8 images at a time and see every prediction with its confidence. Nothing is saved.
- From the dataset or your own upload
- Images and video
- Labels in this image, with confidence
Image recognition software that compares every version.
Each model shows its accuracy and the repository version it learned from. Ask Vero to compare two versions.
- Accuracy per model
- Repo version and size
- “Compare v3 and v4”
Sample dataModel questions.
Do I need machine learning experience?
No. Pick a model type, a folder, a time budget and a size. Or ask Vero and approve its plan.
How many labeled images do I need?
10 to start (2+ classes for classification). Aim for 20 to 50 per class. Anomaly detection (Ultra) needs only good images.
How do I know a model is good?
Each model shows its accuracy in the AI Models panel. Test runs it on up to 8 new images and shows every prediction with its confidence.
What's the difference between Mini and Plus?
Mini is lightweight and runs on CPU. Plus is production-grade and comes with Pro and Ultra. The Extensive time budget is on Ultra.
Is VisionRepo image recognition software?
It trains image recognition models on your own images: an image classification model, object detection, image segmentation or image anomaly detection (Ultra). It works as image segmentation software too, with Brush, Polygon and Smart Polygon for the labels.
Train your first model today.
Free plan, no credit card. Bring your own images or start from a sample.