InCloud AI Case Study | Orient Software

Delivering three AI detection solutions with up to 99.8% accuracy for InCloud

/ AI-POWERED DETECTION · COMPUTER VISION & DEEP LEARNING /

High-accuracy, real-time AI detection across three domains

AI detection solutions delivered

99.8%

peak detection accuracy (5ms speed)

/ SUMMARY /

InCloud is a Japanese technology company specializing in AI-powered software solutions, building applications in computer vision, real-time monitoring, and automation for industries that demand high-accuracy detection and predictive analytics. InCloud partnered with Orient Software to develop three AI-driven detection solutions that enhance real-time monitoring, automation, and accuracy across multiple applications. Using computer vision, deep learning, and mobile AI, the projects tackled three real-world challenges: digital display recognition, breathalyzer digit tracking, and fall detection for elderly individuals.

Location

Japan

Cooperation period

7+ years (ongoing)

Industry

Cloud solutions provider (AI software)

Engagement model

Dedicated development team

Services used

AI development

ML engineering & MLOps

Mobile app development

/ The situation /

Three detection problems, one high bar for accuracy

InCloud builds AI-powered software for industries that can’t tolerate detection errors. It came to Orient Software with three distinct challenges - each demanding real-time performance, high accuracy, and the flexibility to run across online, offline, mobile, and IoT environments.

Digital display recognition

InCloud needed an AI system to detect and digitize 7-segment digital displays for real-time tracking and automation - working both online and offline, at high speed and accuracy.

Breathalyzer digit reading

InCloud sought a mobile AI solution to automatically read and track digital numbers from breathalyzer devices under varying lighting, across mobile and desktop.

Fall detection for the elderly

InCloud needed a thermal-camera AI system to monitor elderly individuals and automatically detect falls in bathrooms in real time, triggering immediate alerts to prevent serious injury.

/ THE SOLUTION /

Purpose-built AI for each environment

Orient Software assembled a dedicated AI team to design and deliver all three detection solutions, drawing on computer vision, deep learning, and mobile AI. Each solution was engineered for its own environment: real-time display recognition, mobile-based digit tracking, and thermal fall detection, all built to InCloud's high bar for accuracy.

AI-powered digital display detection

Orient’s team built a computer-vision detection system using YOLO and multi-box detection, with optimized screen localization for precise character recognition – reaching 99.8% accuracy at a 5ms detection speed for true real-time recognition.

AI-powered breathalyzer digit recognition

The team built a lightweight mobile AI application using SSD models with MobileNet architecture, integrating digital digit recognition for precise breathalyzer data extraction – delivering 98%+ accuracy with smooth, consistent performance across mobile and desktop.

AI-powered thermal fall detection

The team built a fall-detection system using Convolutional Neural Networks (CNNs) for real-time image classification, with custom convolution layers tuned for thermal imaging – achieving 98% accuracy and triggering immediate emergency alerts.

/ THE RESULTS /

High-precision, real-time monitoring across the board

Orient Software’s AI-powered solutions delivered high-precision, real-time monitoring for InCloud’s applications - improving automation, efficiency, and safety across domains.

99.8%

digital display accuracy

The computer-vision system recognizes 7-segment displays at 99.8% accuracy with 5ms detection speed, enabling real-time tracking both online and offline.

98%+

breathalyzer & fall-detection accuracy

The mobile breathalyzer reader and thermal fall-detection system both deliver 98%+ accuracy, with instant alerts that improve safety and responsiveness.