Segmentation
This technology uses AI to classify each object in images at the pixel level.
Each pixel in the image is inferred as to which object it belongs to. It can be used to detect objects that are not suitable for rectangular detection, such as road lanes, cracks in concrete, etc.
USE CASE
Surveillance camera
It identifies people and other objects in the video for person detection and behavior analysis. It can also detect specific scratches and cracks at construction sites to help prevent accidents.
Automatic driving
From the camera image, it classifies and identifies road white lines, sidewalks, buildings, etc., pixel by pixel, and is used for "recognition" by the automatic driving vehicle.
Example of practical application: Crack detection support system by deep learning
In collaboration with Shimizu Corporation, we conducted a demonstration experiment for the construction of a support system aimed at preventing accidents caused by collapsed excavation surfaces near tunnel faces at tunnel excavation sites. For details, please see the .
Medical Image Analysis
Can be utilized for MRI, CT scans, etc., and assists doctors to diagnose diseases by segmenting specific tissues and lesion areas.
APPLICATIONS
IMAGE PROCESSING
Image Noise Reduction
AI technology that reduces high-ISO noise caused by low-light shooting to achieve high image quality
Video Noise Reduction
AI technology to achieve high image resolution in real-time for video recording where long-second exposures are difficult to use
Super Resolution
AI technology to enhance low-resolution digital images while preserving sharpness
Deblurring
AI technology to reduce blur caused by camera shaking or shooting moving subjects
Deraining
AI technology to remove water drops on the surface of the camera lens from the image after shooting
IMAGE RECOGNITION
Object Detection
AI technology to detect where people, vehicles, and other objects are located in an image
Segmentation
AI technology to divide colors by object areas
Pose Estimation
AI technology to estimate the position of human joints and major points (neck, shoulders, elbows, wrists, ankles, etc.) and their connections
Classification
AI technology to recognize objects in images and classify them
Anomaly Detection
Detects anomalies by training only normal images