We have developed an on-orbit object detection algorithm as part of our
“Satellite that Observes Other Satellites” development program.
The algorithm processes stored images, determines whether an object of interest is present, and moves the images to separate directories based on detection results.
Since the logic is specialized for object detection and does not rely on AI models, it can operate with relatively low power consumption on edge devices.
Operation has been verified on platforms such as Raspberry Pi Zero.
The middleware can be used for image recognition in edge environments, and is applicable not only to spacecraft but also to drones, robots, and other edge systems.
Please feel free to contact us if you are interested.
検出のシグナルとして青い四角で囲ってあります
Intended Use Cases
- On-orbit object detection
- Object detection in edge environments such as drones, surveillance systems, and robots
- Image filtering prior to transmission in low-bandwidth environments
Key Features
- No AI models required; no training data needed
- Lightweight, rule-based detection logic
Supported Platforms
- x86 Linux, embedded Linux, and similar environments
- Supported on any platform capable of running C++
- Porting to other programming languages is possible upon request
- Please contact us regarding CPUs embedded within FPGA devices