AI Development｜i-enter Corporation,Ltd.

Image recognition by deep learning AI Development

We utilize image recognition based on deep learning to promote DX in all business and business categories. In i-enter, image recognition can be realized on various platforms such as the Web, smartphones, and edge computers. In addition, we can also support the construction of original neural networks and the construction of PCs dedicated to image analysis according to your needs.

Reduce burden and improve work efficiency with AI

From manufacturing to construction, fisheries, sports, and various manufacturers, we support all industries and business categories.

quality check
Anomaly detection
Intrusion Detection

Posture estimation
Face recognition
High-speed object recognition

Our Patented Technology

April, 2020: Patent granted

Name of the invention: Fish Size Calculation Device

Patent number: Patent 6694039

Application number: Patent application 2018-218922

Discription: A system that enables accurate measurement of the actual size of migrating fish

April 2020 Patent Acquisition Certificate

Detects the body of the fish with image recognition and measures the length and height of the fish from the location information

The size of fish in aquaculture cages involves a lot of manual work of removing and measuring fish directly with a net, which has problems such as errors by measurers, stress on fish, and difficulty in accurately grasping the state of fish growth. Therefore, we have developed a system that utilizes deep learning image recognition to measure the size of fish without touching it.

April 2021: Patent granted

Name of the invention: Posture Estimation Learning System
Posture estimation learning method and data creation program for machine learning

Patent number: Patent 6868875

Application number: Patent 2020-199646

Discription: Generating learning models of posture estimation by using virtual space

Creating learning models of posture estimation all at once in the system without manual inputting

In general, the creation of training data for posture estimation involves a lot of manual work, and it takes time and cost to create a large amount of data. Therefore, we have developed a system that performs everything from generation to learning of training data related to human posture estimation. This system uses 3D models and motion data to capture movements from various angles in 360 degrees and automatically acquires skeletal information.

Virtual Space Attitude Estimation System

Works

Intrusion detection of hazardous areas
Using network cameras and posture estimation technology, we have developed a system that warns workers in real time with alert lamps and buzzer sounds when they enter a dangerous area. By using AI to detect intrusion into dangerous areas, accident prevention and safety improvement are realized without the need for lifeguards.
Non-contact temperature measurement
As a measure to prevent the spread of new coronavirus infection, we have developed a system that measures the temperature without contact and warns when an abnormality is detected. Since it uses face recognition technology and an infrared camera, you can measure your temperature even while wearing a mask. It reduces the risk of contact infection and realizes safe and efficient temperature measurement.
Delivery Form Analysis
Using posture estimation technology, we developed a system that analyzes the delivery (stone-throwing) form of curling players without attaching sensors to the body. Two visible light cameras simultaneously record the delivery form, analyze it with AI, detect the skeletal position of each part, and generate a 3D skeleton.
Stone Tracking
We developed a system that recognizes delivered curling stones with 12 cameras installed on the ceiling, measures their coordinate position, and visualizes the trajectory of the stones. Image recognition detects the infrared rays of stones with built-in infrared LEDs developed in-house, automatically measures the position of the stones, and visualizes the trajectory.
Posture estimation with smartphone
Utilizing deep learning image recognition technology and skeleton estimation technology, we have developed a system that estimates images taken with smartphones in real time. It can be used for exercise form analysis, posture checking, and various situations. By using a smartphone, we can realize easy and familiar skeletal estimation.
Entry/exit management by face recognition
Developed an access control system using face recognition. When entering and leaving, face recognition is performed, the door (electric lock) is unlocked, and entry / exit is recorded. By using this system, there is no need to manage access using IC cards, and it is possible to improve work efficiency and reduce costs.
Management of aquaculture tanks
We are researching and developing a system that analyzes web camera images with AI and automatically monitors the condition of fish and the aquarium environment. Numerical values are recorded from images from multiple digital water temperature gauges through AI analysis. Even in a dark room, it is possible to analyze aeration, water flow, and fish conditions from night vision images.
Streamlining training data creation
We developed an application that photographs objects reproduced in 3DCG and generates training data on Unity. You can batch generate a large amount of training data by importing a CSV file. It was no longer necessary to visit the site for AI quality inspection and manually photograph various parts, and we were able to achieve significant operational improvements.

3DCG Production Achievements

We also support the creation of 3DCG models that can be used to create training data.
We can handle a wide range of products from the creation of small 3D objects to the creation of 3DCG spaces, and we can create models according to your needs.