CORE is an image and video classifier that screens data, images and videos for illegal and inappropriate content. It supports digital forensics, eDiscovery and other review platforms with media analytics as a pure API solution. CORE integrates seamlessly into content review and Trust & Safety workflows.
Benefits of CORE
Best-in-all AI Classifiers CSAM classifier with exceptional precision and recall. Unique classifiers for narcotics, weapons/violence outperforming others. Ability to add novel classifiers quickly and easily for specific use-cases.
Easy integration Media objects are directed to CORE during or after processing in the existing platform, automatically classified based on the chosen AI classifiers, and tagged for review in the integrating platform’s frontend.
Detects 1st Generation CSAM CORE uses a combination of object classifiers and metaclassifiers for CSAM detection. Trained on 3rd party CSAM data using image encodings. True positive detection rate over 80% with a false positive rate under 5% .Can be continuously retrained and improved using new encodings
Natural Language Searching CORE can be enabled to search for picture contents by describing them with natural language (text2image), or by providing a picture and searching for contextual similar images (image2image).
Fully Containorised Allows for easy adoption on-premise or into a common cloud + edge environments. Due to the modular containerized setup, CORE in the cloud can be scaled horizontally or vertically.
Additional Features Multi-lingual OCR, hash & PhotoDNA matching, face recognition from uploaded reference images, 2D pattern recognition (logos, art, bank notes, tattoos etc.)
System Requirements
For on-prem installation
GPU: Nvidia Quadro RTX 4000 or equivalent Nvidia GPU