top of page
빈 공장

AI Powered Multi-Camera
Tracking Solution

Object detection and location analysis solution

Cai LOGO-

Multi-Camera Tracking

CAI revolutionizes spatial information through AI technology.

CAI_Location

Utilizing existing CCTV without additional markers or sensors,
our solution employs deep learning to achieve real-time location measurement and object detection.


By analyzing the environment, it recognizes objects within the
space and measures distances between individual entities.

To enhance accuracy and repeatability,
we continuously gather and train with learning data, elevating engine performance.

AI Feature

Diverse angles, elimination of blind spots, partial body estimation analysis

Storing personal information

in an anonymized manner

Quantifying location information for real-time integration with external systems

CAI-Locaiton

AI Powered Multicamera
Tracking Solution
for Smart Factory

CAI_Location

Embedded AI Computing Engine

CAI-Locaiton

Built-in Alarm Speaker and Lighting

CAI-Locaiton

350-degree Pan, 120-degree Tilt functionality

No need for a separate AI server setup

Alarm messages and warning lights emitted in case of anomalies

6 times larger detection area compared to fixed cameras

CAI_Location
CAI_Location

Build Smart Factory 

Cai Screen

CAI transforms spatial information through AI technology.

It detects and reports the real-time positions and situations of workers,

hazardous materials, and moving equipment.

No special markers or sensors are required.

Existing CCTV infrastructure can be seamlessly

upgraded to intelligent AI cameras.

01

Trained with 600,000+ real-world training data, it supports world-class recognition performance. As a result, even when only parts of a worker's body are visible, it can recognize them, addressing blind spots and on-site situation issues.

World-class Recognition Performance

02

By exponentially expanding the number of cameras, you can eliminate blind spots in the area and enhance object tracking accuracy.

Unlimited Space Measurement

03

Generating additional training data to adapt to daily changing on-site situations, we maintain optimal conditions and accuracy.

Real-time Additional Learning

04

Within 0.1 seconds, it identifies an object's position and situation, with location measurement accuracy within a 10 cm range.

Swift and Accurate Positioning

CAI system instructions

Multi-camera Analysis

You can simultaneously connect from 1 to a maximum of 16 cameras in parallel.

It recognizes tracking subjects as a single object,

enabling comprehensive area analysis without blind spots.

[ENG]파이미디어랩_IFA Leaflet_v.99.png
작업자의 구역 식별

Identification of worker’s zone

위험구역 진입 시 경고

Measurement of distance between workers and objects

작업자 동선에 따른 히트맵

Alerts upon entering hazardous areas

신체 일부분으로 인식/판별

Recognition/discrimination using body parts

작업자 동선 추적

Tracking of worker movement paths

작업자와 물체 사이 거리 측정

Heatmaps based on worker movement paths

The Competitive Edge of Cai Location

The technology developed by PAI Media Lab

addresses obstacles that are challenging for a single camera to detect,

as well as the ability to detect objects within

complex structures andaccurately position them.

This is achieved through IPM (Inverse Perspective Mapping) and Calibration techniques,

enabling the transformation of fragmented images acquired from camerasinto a panoramic view. This patented technology models the area of interest, allowing precise measurements.

[ENG]파이미디어랩_IFA Leaflet_v.99.png

Benefits of Adoption

Icon

Affordable Initial Implementation Cost

No need for costly AI server setup, reducing initial expenses by 1/3

Icon

Easy Installation

and Operation

Supports dedicated interfaces for wireless network setup and monitoring

Icon

Various

Reporting Features

Reports statistics data such as personnel location, heatmaps, people counting, and weekly/monthly trends

System Demo

Real-time streaming using general FHD-level CCTV to analyze the target and measure the location

(average analysis speed 0.01 sec, measurement error 0.1 m, measurement range, 12 m x 10 m)

Wildlife Detection AI Algorithm

Wild animal recognition deep learning algorithm module based on R-CNN

Coordinates, relative distance, and heat map reporting for each recognized object

Recognize semantic segmentation and classify objects

- Minimize false positives by applying Denoising / Background Subtraction technology

야생동물 인식

# Real-time detection in wildlife recognition system using our engine screen

Solution Application History

PARTNER'S LOGO
PARTNER'S LOGO
PARTNER'S LOGO
hyundai

KOREA ELECTIR POWER CORPORATION 

Object Recognition Algorithm within the Safety Reporting App  / 2020.09

KOREA Railroad

Wildlife detection AI algorithm  / 2021.03 ~

KT

Smart Building Disaster Prevention Algorithm  / 2021.07 ~

HYUNDAI MOTOR GROUP

Worker Location Tracking Algorithm / 2021.09 ~

 

Solution inquiry: bsyeo@peoplei.kr

​Fi Media Lab Co., Ltd.

2F, 597 Hoam-ro, Gwanak-gu

Business registration number 156-86-01908

bottom of page