Steel optical recognition & tagging (SORT) technology
SORT leverages advanced computer vision and multimodal language vision models to classify and track scrap metal in real-time, enhancing steel yield management through a network of sensors. It offers customizable solutions using open models trained on large datasets.
Overview
Steel Optical Recognition & Tagging (SORT) is an innovative solution utilizing real-time computer vision systems to identify, classify, and track scrap metal across a network of sensors. By employing multimodal language vision models (LVMs) that are trained on extensive datasets via large-scale GPU clusters, SORT enhances steel yield management. This system allows for third-party customization to adapt the open models to specific needs in computer-vision-driven yield management, ensuring flexibility and precision.
Technical specifications
Key features:
- Utilizes advanced multimodal language vision models for accurate material classification.
- Supports real-time processing with a throughput of 20 frames per second.
- Incorporates a vector database for anomaly detection and model drift identification, enhancing system reliability.
- Offers a customizable architecture that reuses existing components and develops new ones as needed, including annotated datasets.
Technology readiness level
Currently at Technology Readiness Level 4, SORT has undergone initial validation in controlled environments. The future development plan includes a three-phase approach: designing a custom solution architecture, implementing data, model training, and inference pipelines, and conducting field trials to refine and validate the system in real-world conditions.
About Aegean AI Inc
Aegean AI Inc is a research and software development company focused on creating Physical AI agents that perceive, reason, and act in real-time. By integrating computer vision, natural language processing, and robotics, the company develops systems capable of interpreting complex environments through multimodal inputs. Their core offerings include the Auraison platform, a multimodal agent runtime designed to support mission-critical industries, and various technology demonstrators that showcase applications in manufacturing quality control, remote sensing, and sports analysis. Their approach leverages deep learning, reinforcement learning, and advanced sensor data processing to enable machines to execute tasks based on natural language instructions.
The company serves organizations and mission-critical workers by addressing challenges in industrial automation, geographical information systems, and operational efficiency. By enabling interactive, real-time AI agents, Aegean AI helps businesses reduce operational silos and improve communication between disparate vendor systems. Founded in 2020 by CEO Pantelis Monogioudis, the firm draws on extensive academic and industry experience in AI, telecommunications, and perception to provide specialized consulting and software solutions that improve automated decision-making and visual understanding in real-world environments.
delivered to your inbox.