One-click installation
No Docker or complex setup.
Local data preparation and model training
Teach AI to recognize what matters.
Use your own images or videos to create custom AI without coding or complicated setup.
✓ One-click installer for Windows, macOS and Linux. No Docker required.
AI auto-annotation
Annotate images and videos faster.
Your data stays local
Keep your projects on your machine.
From raw data to model
Three straightforward stages.
1
Open your data
Import images or video and define the objects or conditions that matter.
2
Review AI-assisted labels
Use assisted and automatic annotation to prepare more data with less repetitive drawing.
3
Train and export
Select a supported model, train locally or in Colab, review results, and export the model.
Training · RF-DETR Running
mAP50 0.842Precision 0.932Recall 0.813
Benefits
Spend less time managing the machine-learning process.
Reduce repetitive labeling
Use SAM-assisted tools to create boxes, polygons, and masks faster.
Keep projects organized
Manage images, video, classes, annotations, datasets, and models in one app.
Avoid environment setup
Install the application without assembling containers or development environments.
Train where it makes sense
Use your compatible GPU or send heavier jobs to your Google Colab session.
Keep ownership of the result
Export annotations and models in supported standard formats.
Work privately
Keep proprietary images, annotations, and models on your own computer.
ImagesAnnotationsDatasetsModelsLocal training
When to use VisoLabel
Use a custom model only when the use case requires one.
VisoLabel is useful when a general model cannot reliably recognize:
Your own objects
Build a dataset around what is specific to your operation.
- Your particular product or component
- A company-specific defect
- Specialized vehicle or equipment categories
Your own conditions
Train for categories and environments that general models do not cover.
- Unusual objects or materials
- Conditions from your camera environment
- Classes unavailable in existing models
◉ Once the model is ready, move it into VisoNode or another supported deployment environment.
FAQ
Questions teams usually ask
?Does my data leave my computer?⌄
Local annotation, dataset management, and supported local training workflows remain on your machine. Data leaves the computer only when you explicitly choose an external training option such as your own Google Colab session.
?Do I need an NVIDIA GPU?⌄
A compatible GPU is useful for faster AI annotation and local training. Basic annotation does not require one, and compatible cloud training can be used for heavier jobs.
?Do I need machine-learning experience?⌄
VisoLabel removes much of the environment and toolchain setup. You should still understand the purpose of your classes and evaluate whether the trained model performs adequately for your application.
?Can I export my work?⌄
Supported annotation and model formats can be exported for use outside Pysource.
?Can I use the model with VisoNode?⌄
Supported models trained with VisoLabel can be used as part of a VisoNode workflow.
Try VisoLabel on your own data.
Install the free version, open a project, and see how much of the annotation process can be simplified.