Ximilar - Remove Background Solution

Ximilar - Remove Background Solution

Ximilar currently uses OpenVINO for accelerating DL inference on object classification models. In order to extend the deployment portfolio offering more solutions to the customers, Ximilar is including the retail image content management - background removal DL model (based on U2-Net). This background model optimized with OpenVINO is used in three different ways: As a standalone service that can be used by retailers for changing or removing background from product images; as an image augmentation technique for Custom Recognition Service, which helps us and our customers expand their training dataset and improve the overall robustness and accuracy of the model; and as an image augmentation technique for Custom Similarity Service. It is used for changing background and generating pseudo-real photos for training the model. This helps the search technology achieve the best recall performance.

*Please note that member solutions are often customizable to meet the needs of individual enterprise end users.



  • Foreground-Background Separation Model



WorldwideAI Software/SaaSCross-IndustryRetailImage/Object Detection/Recognition/ClassificationIntel® Xeon® Scalable ProcessorIntel® Distribution of OpenVINO™ Toolkit powered by oneAPIIntel® Optimization for TensorFlow*Models can be trained - data input only requiredLinux UnetDeep Learning