オービタルネット List of Products and Services

■Product Group ・Web-based Fixed Asset Management Support System

• CRM (Customer Relationship Management System)
• Store Analysis System
• others

This is an object counting system that combines deep learning object detection technology with object tracking algorithms. It tracks vehicles from footage captured by a fixed camera and counts the number of cars that pass through a specific section. It is necessary to train the system on the types of vehicles to be classified as needed.

• Big Data
• Database
• Analysis and prediction system

This is an object counting system that combines deep learning object detection technology with object tracking algorithms. It tracks walking individuals from footage captured by a fixed camera and counts the number of people who pass through a specific section. Unlike traffic vehicles, where predicting movement is relatively easy, human movement is complex, and long-term tracking is extremely challenging. Particularly at crosswalks where individuals moving in different directions intersect, technology that focuses on specific areas to detect and count pedestrians is required.

• Consulting companies
• Analysis and prediction system
• Other measuring instruments

Real-Time Geo Tracer is a feature detection and geo-coding tool that utilizes the instance segmentation technology "Mask_R-CNN." It extracts specific features from a large volume of aerial photographs and records them as geospatial data. In addition to feature extraction, it traces the boundaries of features to create polygons, allowing for more accurate understanding of feature characteristics such as area calculation. * Mask_R-CNN is a technology that performs class classification and segmentation for each detected object individually.

• Big Data
• Solar power generation
• Consulting companies

The Real-Time Geo Detector is an aerial photo land feature interpretation system that applies deep learning real-time object recognition technology, which simultaneously identifies candidate regions and classifications from images and videos. It is realized using the CNN object detection algorithm YOLOv3 and the open-source deep learning framework Darknet. Although this technology is originally intended for real-time object detection from videos, it leverages the characteristics of real-time object detection algorithms to instantly detect land features from large volumes of aerial photographs instead of videos. By operating on a GPU-equipped PC, it can detect land features from wide-area aerial photographs at ultra-high speed and register their coordinate information in a spatial database.

• Big Data
• Solar power generation
• Consulting companies