Train and curate
Prepare image datasets, review quality findings, annotate, and train classification, detection, or segmentation models.
Read SDK documentation →Capabilities
Tools for the whole application
The SDK supports both ready-made model workflows and custom development.
Run and compose
Use model inference with video, tracking, counting, region timing, and configurable ordered pipelines.
Explore runtime tools →Deploy on edge hardware
Run compatible ONNX models and compile supported detection or segmentation checkpoints for Hailo targets.
Explore deployment →Studio and SDK
Move from visual workflow to Python code
Vision Studio is the desktop workbench for dataset review, annotation, spectral analysis, training, and pipeline previews. The SDK gives developers Python APIs and a command-line interface for custom applications.
Install the package with pip install -U abraia. Some models and hardware targets need optional dependencies. Studio also works with multispectral and hyperspectral scenes; explore multispectral imaging solutions.
Face recognition
An SDK capability for specific applications
The SDK includes face representation and identification for matching faces against a reference set in images or video. The application still needs enrollment, evaluation on representative footage, and integration designed for its setting.
See the current SDK documentation and examples for API details. For people counting without identification, see queue monitoring.
Bring your vision problem to us
Tell us about the images, sensor, operating environment, and decision you want the system to support. We can identify a practical first evaluation.