IMAGE RECOGNITION & TRACKING

Our AI image recognition & tracking technologies provide economic, flexible, robust and expandable services that can be operated with existing hardware platforms

In image recognition & tracking services, the processing tasks usually require huge GPU based server usage that results in high operational cost and extra maintenance. In addition, when such services are deployed in the CPU or low-level GPU based edge devices without proper AI model design, these often result in errors, low-accuracy, and or delayed response as they cannot interact seamlessly with high-computational nodes.


At DDE, we have developed our own camera-based image recognition and deep learning model unification technologies that can operate with existing hardware platforms without requiring any additional sensors to provide the most economic and powerful AI models. With our design, the AI and technologies are ensured to be flexible, robust and expandable even operating under a very low hardware condition.

IMAGE RECOGNITION & TRACKING

Our AI image recognition & tracking technologies provide economic, flexible, robust and expandable services that can be operated with existing hardware platforms

In image recognition & tracking services, the processing tasks usually require huge GPU based server usage that results in high operational cost and extra maintenance. In addition, when such services are deployed in the CPU or low-level GPU based edge devices without proper AI model design, these often result in errors, low-accuracy, and or delayed response as they cannot interact seamlessly with high-computational nodes.


At DDE, we have developed our own camera-based image recognition and deep learning model unification technologies that can operate with existing hardware platforms without requiring any additional sensors to provide the most economic and powerful AI models. With our design, the AI and technologies are ensured to be flexible, robust and expandable even operating under a very low hardware condition.

USE CASES

DDE's biometric recognition, object recognition and tracking technologies are used in a wide range of industries from fine parts production to retail services

For easy and efficient inventory management for on/offline bookstores, DDE has developed book and spine image recognition algorithms from book and library images, customised algorithms and system models that can detect and identify books from images by extracting each text information via OCR and each non-text information with estimation algorithms that are then mapped with massive book information DB for inventory.

With our newest book spine detection system, users can easily participate in network inventory by taking pictures of their bookshelves and uploading them via their cell phones. The AI then identifies and catalogues the inventory from the uploaded images.

DDE’s multi-class defect inspection system combines and reduces existing DL models into unified models, reducing GPU usage by more than 90% on average as well as reduction in server and maintenance costs. Our defect inspection system is not only cost-efficient, but also highly accurate – having achieved 99.9% accuracy for our latest project.

Based on our highly precise camera recognition technologies, DDE’s camera-based unmanned store system detects everything in its view, including but not limited to people, pose, hand, product, basket, different motions of picking up and dropping of inventory, etc. Our unmanned store operating system can be operated with only a small number of cameras without any other sensors or equipment, providing an economic and easy business model for retail professionals.

Originally developed to monitor fire escape safety distance in real-time, DDE’s Auto Distance and Route Calculation AI uses our object detection and image processing technologies to extract depth value from 2D images. Our Auto Distance and Route Calculation AI can be used with existing 2D security cameras to be further customised for client’s inventory and security system support, and can be applied to track objects and human movements and to send alerts when there is a security risk as part of security management.

DDE’s facial recognition technologies offer an economical and reliable solution for fast and accurate facial identification. Developed with the pandemic in mind, our facial recognition AI recognizes facial features even when wearing masks within milliseconds. Developed using an advanced CPU-based model, instead of a GPU-based one, our facial recognition AI is economical and can be deployed with any existing cameras, equipment and systems.

Unlike rule-based self-driving systems, which require processing of numerous parameters for various environmental changes when modeling, that result in endless coding, DDE’s learning-based self-driving AI clones human driving behaviours in response to various road and environmental conditions with easily available driver behaviour data (i.e., steering angle, steering angular speed, velocity, etc.) from ECUs of existing vehicles and visual data from low-priced webcams.

Superior in both cost efficiency and performance, our self-driving AI does not require LiDAR or any other sensors and can operate in off-road and harsh weather conditions.

Data Design Engineering

Asia HQ

Addr. 2F, 12, Digital-ro 31-gil, Guro-gu, Seoul, South Korea

Tel. 070-8830-1621 | Email. info@datadesign.engineering


© Data Design Engineering all rights reserved.

다.