SYNTHETIC IMAGERY
Synthetic Imagery as a Service
PRENAV’s Synthetic Imagery as a Service program augments your existing AI/ML training datasets by dramatically increasing the size and diversity of your training data. After you provide PRENAV with a 3D model of a part and a few photos of example defects, PRENAV will utilize our proprietary Visual Effects (VFX) tools, expertise, and image processing pipelines to render thousands of synthesized and annotated images for you each month. These images will match your production inspection camera specifications (pixels, focal length) and the possible lighting environments for the scene, and will include both good condition examples and defective examples.
PROCEDURALLY GENERATED IMAGES FOR DEEP LEARNING TRAINING
Each defect will be designed and customized for your parts, and can be procedurally generated through a variety of tunable parameters affecting the defect’s size, geometry, location on the part, textures, and more. Synthetic images can be generated with a variety of virtual backgrounds and from a variety of camera angles and proximities, or certain variables can be fixed to match your production inspection camera setup.
Through our Synthetic Imagery as a Service program, PRENAV continuously provides you with either brand new or enhanced synthetic data (based on your real-world feedback) to improve the accuracy and recall of your existing AI models.
If you’re interested in applying Synthetic Imagery into your deep learning training process, please contact us for pricing. We offer a variety of options to get you started.