スペクトラム・テクノロジー List of Products and Services

This is an all-in-one kit that utilizes Raspberry Pi 5 hardware and AI agents provided by AWS cloud services, such as AWS Bedrock. It operates in Python and offers multi-agent capabilities through agents, knowledge bases, and RAG. It leverages customer data and provides numerous practical examples (open source). It can be widely used by everyone from students to developers.

• Software (middle, driver, security, etc.)

This kit consists of Jetson Orin Nano, Livox Lidar, a camera, etc., and provides pre-installed software and manuals. It can be used for 3D mapping (Livox_mapping, Lio-Livox, Fast-lio). It also supports video-based 3D mapping: Fast-livo (LiDAR-Inertial-Visual odometry) from calibration to utilization. Numerous demos are available, allowing for a wide range of applications from basic to advanced. It is compatible with Livox and ROS.

• Image Processing Software

For wandering prevention using Apple products and smart home devices, outdoors, we utilize location tracking via Apple Watch or AirTag, and communication (only with Apple Watch). Indoors, we use smart home products (from Wulian) such as door sensors, motion sensors, and emergency notification buttons. Management can be done collectively through an iPhone app, allowing for immediate awareness of wandering incidents (door sensors) and locations (location tracking), enabling prompt protection even from a distance. Additionally, there is no need to impose restrictions (such as prohibiting going out) on individuals with dementia, reducing the burden on both parties. We hope you take this opportunity to make use of it.

• Other information systems

The Beginner's Image and Video Generation AI Learning and Development Kit 2 (Ubuntu version) provides open-source software (such as Stable Diffusion, Runway, Copilot, DeepSeek, etc.). It allows for a wide range of learning and application from the basics to advanced techniques in image generation (text to image, image to image) and video generation (image to video). It is particularly effective for those in the game and entertainment industry for content creation and editing as a means of increasing efficiency. Additionally, for beginners, installation and services are provided.

• Software (middle, driver, security, etc.)

The first image and video generation AI learning and development kit will be sold in February 2024, allowing users to utilize image generation (from text to image, from image to image) and video generation (from image to video) from the basics to advanced applications. This time, we will provide an all-in-one version using Jetson Orin Nano, aimed at beginners, covering everything from hardware to software. No installation is required, and you can immediately start learning and developing image generation AI (video generation is not included). It is recommended for those who have not experienced this before. We will provide open-source software (such as Stable Diffusion, Copilot, etc.), enabling a wide range of learning and application from the basics to advanced uses of image generation (from text to image, from image to image).

• Image Processing Software

This kit consists of Jetson Orin Nano, IMU, Livox Lidar, etc., and provides pre-installed software and manuals. It can be used for 3D mapping (Lio-sam, lidarslam_ros2, Cloudcompare) and for drones. We also offer numerous demos, allowing for a wide range of applications from basic to advanced. It is designed for beginners and was released in August 2024 as the LiDAR Development Kit 2.

• 3D measuring device

OpenUSD (Universal Scene Description) was developed by Pixar and established an alliance in 2023 in collaboration with Nvidia, Apple, Autodesk, and others. It will become a powerful tool for next-generation 3D content creation, enabling a common language, collaboration, and reuse. It can be used for animation, metaverse, and digital twin simulations. NIM integrated with generative AI is also provided. It can be utilized by companies and schools producing next-generation 3D content.

• Image Processing Software

The first KVM virtualization introduction kit uses Ubuntu and macOS as the host OS, and supports Windows, Unix, Linux, and macOS as guest OS. KVM is adopted by major cloud providers, is open source, and can be used free of charge. Support relies on the community. We recommend utilizing KVM for its excellent cost-performance ratio. This introduction kit provides image files for major guest OS and configuration manuals, as well as support, so you can use it with confidence.

• Other server related

This kit consists of Ubuntu 22, GPU, and Livox Lidar, and provides software and manuals. It offers software and applications that can be utilized for 3D mapping (Lio-sam, fast lio ros2, Cloudcompare), autonomous driving (autoware, Scenario sim, awsim), and drones. Numerous demos are also provided, allowing for a wide range of applications from basic to advanced. This is the ROS2 version of the LiDAR development kit that was sold in October 2022. It is intended for intermediate to advanced users as it requires installation from source.

• 3D measuring device

This product allows you to simply place our smart home leak sensor at the pool's overflow drain, and it will notify your smartphone. It can be easily shared with multiple people, and there are no monthly fees. You can implement immediate measures without renovating existing facilities. Please make use of it.

• Other network tools

The 2024 issue has raised concerns about a shortage of drivers, but there were 310,000 car accidents (31,000 in 2023) and 2,618 fatal accidents (2,618 in 2023), with an estimated 20% of those being caused by drowsy driving (60,000 cases). Drowsy driving warning devices in trucks and other vehicles are found in less than 0.1% (about 200,000 out of 14 million trucks) and are not widely adopted. This time, we will be selling a kit to prevent drowsy driving accidents before they occur.

• Other network tools

This kit consists of a Raspberry Pi 4B, IQaudio DAC+, and open-source moode, among others. It supports high-resolution playback and recording, and can be operated using a PC, smartphone, or tablet browser. The setup and operation are simple, making it ideal for beginners. Additionally, it can connect to Bluetooth speakers and can be used as a DLNA server or NAS. It is compact and lightweight, making it easy to connect with your existing record player and audio amplifier. It is recommended as an introduction to high-resolution audio.

• Other measurement, recording and measuring instruments

Object and person recognition is expected to expand to 68 billion yen by 2027 (Deloitte Tohmatsu Research, September 2023) in areas such as crime prevention and security, defect detection, and AR. Currently, there are many AI software options for object and person recognition, but few are simple and capable of high-speed processing. The Mediapipe developed by Google is open-source and provides software that can be implemented from smartphones to the web. It enables high-speed processing without a GPU using the latest Raspberry Pi 5. It is suitable for users from beginners to advanced, making it ideal for those learning and developing object and person recognition.

• Image analysis software

This kit provides open-source software (such as stable diffusion, runway, copilot, etc.). It allows for a wide range of learning and application from the basics to advanced techniques in image generation (text to image, image to image) and video generation (image to video). Each software is available in both local and cloud versions. It is particularly effective for those in the game and entertainment industry for content creation and editing, as it enhances efficiency. Additionally, for beginners, installation and services are provided.

• Other information systems

This kit adds a GNSS-RTK option to the previously released operation management development kit. It incorporates GNSS-RTK with the ublox ZED-F9R built-in, mounted on a Raspberry Pi. The horizontal accuracy is 20 cm, which is significantly better than the conventional several tens of meters. RTK is achieved through the NTRIP method. Data obtained from GPS terminals is connected and stored in the AWS cloud via MQTT, and displayed on a map in real-time. It can be widely used for operation management, autonomous driving, and 3D map acquisition using drones. The price is also affordable, and it can be set up in about a week, ready for immediate use.

• Other network tools

The 2024 issue has raised concerns about the shortage of drivers. This time, we will provide an operation management development kit that can manage driver work schedules and driving skills. Additionally, it can be applied to the management of operations for ships, trains, and in the future, remote management of autonomous vehicles and delivery notifications for drones. We offer a total solution from GPS terminals to web applications, contributing to cost reduction through a pay-as-you-go (cloud) model.

• Other operation management software

Generative AI is expected to be increasingly utilized in society, contributing to improved productivity for companies and enhanced capabilities for individuals. On the other hand, there are challenges such as the inclusion of misinformation in output results and the risk of personal information leakage. This time, we are providing a development kit that uses the API of ChatGPT, a representative of generative AI. It is an all-in-one type development kit with excellent cost performance, allowing both professionals and students to acquire skills and engage in development.

• Other operation management software

This kit consists of a Jetson Nano and a ZED camera, providing hardware, software, and manuals. It offers software and applications that can be used for distance measurement, object recognition, 3D mapping, game development, and scientific calculations. With the all-in-one type provided this time, no software installation is required, making it easy for beginners to use immediately. Numerous demos are also provided, allowing for a wide range of applications from basic to advanced.

• Distance measuring device

In September 2022, WiFi 6E (6 GHz band) became available for use in Japan, and routers, smartphones, USB devices, and more have been released by various companies. We are now starting the sale of the WiFi protocol analyzer V2, which also supports 6 GHz. It will be an affordable and easy-to-use device utilizing a Raspberry Pi and a WiFi 6E USB. It will be provided with the familiar Wireshark, making it widely usable for beginners to advanced users for troubleshooting, performance verification, and more.

• Wireless LAN

LiDAR (light detection and ranging) will be used in many applications in the future, including autonomous driving, surveying, drones, and 3D scanning. This time, we are providing a development kit that allows the creation of 3D maps using LiDAR from Livox, a subsidiary of DJI, with the Jetson Nano (made by Nvidia). It is an all-in-one type, enabling both professionals and students to learn and develop technology using the Jetson Nano and the cost-effective Livox LiDAR.

• Other measurement, recording and measuring instruments

Features - Wireless and easy to place - Covers 30-40m from the gateway - Notifications to smartphone, shareable with members Specifications - Compatible with Zigbee - Detects water levels of 1mm or more - Operates on battery (CR2: sold separately) - Dimensions: 69x29mm (DxH) - Weight: 100g

• Other environmental analysis equipment

LiDAR (light detection and ranging) will be utilized in many applications in the future, including autonomous driving, surveying, drones, and 3D scanning. This time, we are providing a development kit using the LiDAR from Livox, a subsidiary of DJI, which can be applied to autonomous driving, 3D mapping, and drones. With Livox's cost-effective LiDAR, both professionals and students can acquire skills and develop technology.

• Other image-related equipment

DeepStream provides a multi-platform, scalable TLS encryption security that can be deployed on-premises, at the edge, and in the cloud. It has numerous applications, including surveillance cameras, autonomous driving, video distribution, and video analysis. The platform uses Ubuntu and is intended for intermediate to advanced users. Additionally, installation services for beginners are also available. It is recommended to use it in conjunction with the separately sold transfer learning kit (TAO version). A low-code version of the graph composer is also available for beginners to use with confidence. DeepStream is a streaming analysis toolkit for AI-based multi-sensor processing and understanding of video, audio, and images.

• Image analysis software

Using TAO, you can either deploy NVIDIA's operational quality pre-trained models as they are or fine-tune the models to suit various use cases such as video AI and conversational AI. TAO is a core component of the TAO (Train, Adapt and Optimize) platform, which features a UI-based guided workflow for AI creation. The platform uses Ubuntu and Docker and is aimed at intermediate to advanced users. Additionally, installation services are available for beginners. TAO is a tool developed by NVIDIA, which stands for Train, Adapt and Optimize. It is an AI model adaptation platform that simplifies and accelerates the development of AI applications and services for businesses. By fine-tuning pre-trained models with custom data through a UI-based guided workflow, the creation of high-accuracy computer vision, speech, and language understanding models, which previously took months, can be completed in just a few hours. No extensive training or expertise in AI is required.

• Image analysis software

4K live streaming services are increasingly in demand due to the high quality of video and the sense of presence they provide. Until now, they were expensive and required specialized providers and dedicated equipment, but now, using AWS CDN (Content Delivery Network) and laptops, it has become possible for general customers to set them up. We provide the know-how and support for this setup. Please utilize it for event broadcasting, CEO speeches, remote classes, game streaming, broadcasting materials, and more. Equipment is available for rental. It includes features such as quality assurance, access control, and global compatibility.

• Other services

You can comprehensively experience and practice mainstream AI languages such as TensorFlow (developed by Google) and PyTorch (developed by Facebook) from Python, covering everything from learning to demonstrations. In particular, we have added a learning module for high school teachers. This material fully complies with the teacher training curriculum for the high school information course "Information 1," allowing you to acquire and execute programs related to the basics, applications, and simulations of computers and programming in Chapter 3. Additionally, for full-scale use, this program can be utilized by equipping platforms like Windows and Ubuntu with GPUs. Installation work is also available as an option.

• Other network tools

The TensorRT learning and development kit allows for a wide range of skills to be acquired, from training basic models such as MNIST, NMT, RCNN, and SSD using programming languages like Python and C++, to applications such as object recognition and object classification. It can accelerate models created with TensorFlow and PyTorch by up to 40 times, enabling the deployment of these models to edge devices for use in autonomous driving and image recognition.

• Image Processing Software

Content: Kaggle is a platform related to predictive modeling and analytical methods where companies and researchers post data, and statisticians and data analysts from around the world compete to find the optimal models. Prizes such as $100,000 are set, and AI engineers from around the world propose solutions to challenges. Additionally, it offers foundational learning in Python and machine learning, and allows referencing other people's programs. This time, we provide a manual in Japanese for its usage, updated to the latest version (November 2021), free of charge. Features (1) Peace of mind with a Japanese manual The setup method, how to create Notebooks, and case studies are explained in Japanese, allowing you to use it with confidence. (2) Case studies provided Examples using MNIST and Titanic data for CNN and linear models are presented, enabling foundational learning. (3) Explanation of corporate participation The manual also explains how companies can participate in competitions. (4) Provided free of charge To foster the development of AI engineers, the manual is offered free of charge.

• Contract Analysis

Device Defender consists of two components: security auditing of devices and vulnerability detection. Audits are conducted automatically on a regular basis, and measures can be implemented automatically based on the results. Anomaly detection can be configured using either machine learning (ML) or rule-based settings. All features are designed to require no human intervention. The cost is included in the AWS IoT development kit (Pi version).

• Other core systems

AI adoption in companies is still around 20% (Note 1). In the future, it will be widely utilized in areas such as DX promotion, marketing, customer support, and indirect departments, contributing to the resolution of labor shortages. We provide installation services for AI computers, which are complex and have a high level of difficulty, aimed at companies in the consideration stage of adoption. (Note 1) Status of AI and IoT adoption in Japanese companies (2020 White Paper on Information and Communications) The first AI computer will be equipped with AI hardware (GPU) and software (such as TensorFlow, PyTorch, OpenCV, Python, etc.) installed and configured on your existing Windows PC, offered in an environmentally friendly and cost-effective manner. This saves time for intermediate to advanced users, who would typically spend about a week on setup. The supported operating systems are Windows 10, Windows 11, and Ubuntu.

• Industrial PCs

Spectrum Technology Co., Ltd. has started selling the "AI Program Learning Kit (ChromeOS version)." This learning kit is designed for Chromebooks and provides the opportunity to learn AI programming. It covers a range of topics from basics to image recognition, including Python, TensorFlow (including the JavaScript version), PyTorch, and YOLO, as well as demo experience examples (sketching, music, games) in an all-in-one format. Numerous case studies are included, making it accessible for beginners to advanced users.

• Other information systems

This tool (Note 1) is designed to use Raspberry Pi 4B and Kali for port scanning, web vulnerability assessment, password cracking, and hacking tools, and can be used as a training tool for developing white hat hackers through hands-on experience. Additionally, it allows for in-house vulnerability testing on company devices, enabling early response to information leaks, risk reduction, and cost savings. Furthermore, ransomware countermeasures and forensics have been added this time. (Note 1) Using this tool on devices from other companies may constitute a crime, so please use it only within your own company.

• Other security

Using a voltage and current sensor along with a Raspberry Pi, we monitor the battery voltage of a portable power source and send a shutdown signal to the computer in advance if it drops, ensuring a safe stop. Battery voltage and current data are continuously sent to AWS IoT and displayed in real-time on a graph. By reducing peak power during the day and charging with solar panels, we can decrease CO2 emissions. According to our company's estimates, we can reduce CO2 emissions in households by 3% (80 kg CO2) and save on electricity costs: 4,550 yen/year.

• Other energy equipment

The Face and Object Recognition AI Development Kit TF2 can perform full-scale face and object recognition using TensorFlow v2 and a GPU. It is easy to create customized models and utilize pre-trained models for various applications. It can be widely used from schools to businesses. We also provide two types of annotation tools. You can prepare photos and create training files with simple operations. Examples of applications include drowsy driving detection and pose detection. While customers need to install GPU, TensorFlow, OpenCV, etc., we also offer installation services for those who are not familiar with the process.

• Other information systems
• Image Processing Software

Using sensors and Raspberry Pi 3B Plus, you can experience connecting to AWS (Amazon Web Services) IoT, sending emails, and real-time data display in just one week. By using Node-RED, you can significantly shorten the development period. This is recommended for those who are using AWS IoT for the first time or are considering its implementation in the future.

• Other information systems

It consists of a Raspberry Pi 4B, a LiDAR (TFmini-S), and a rotating platform (HAT for Pi), and operates using a Python program. It functions as a 3D scanner using only the LiDAR. The acquired point cloud data (PCD: point cloud data) can be processed into mesh data using MeshLab and can also be output to a 3D printer. It can be utilized from basic to advanced applications.

• Distance measuring device

Since the launch of the AI program learning kit in July 2017, we have received great acclaim from customers such as universities and research institutions. This time, as the fifth generation, we will sell AI program learning kit 5, which allows users to learn the basics of AI in an all-in-one package, incorporating TensorFlow v2 and PyTorch. It covers the latest version of TensorFlow (developed by Google) and PyTorch (developed by Facebook), enabling a wide range of experiences and practical applications from learning to demonstrations. Additionally, we will continue to offer version 1 of TensorFlow, along with explanations on how to transition to version 2. Furthermore, for serious use, this program can be utilized by equipping platforms like Windows and Ubuntu with GPUs. Installation services will also be available as an option. We have released a dedicated learning kit that allows users to learn TensorFlow Lite using the Google Coral (USB accelerator).

• Other information systems

We provide a "Wireless LAN Visualization Service" using a wireless visualization tool (Airmagnet) to clarify the reception area, interference, data communication speed, and the presence of unauthorized devices. By visualizing radio waves, we can determine whether it is necessary to redesign or add wireless AP channels, and we can identify interference devices, allowing us to implement current status assessments and improvement proposals. Additionally, before installation or addition, we conduct surveys using temporary access points in a manner close to real-world conditions. 【Service Overview】 ■ Visualization of Security: Security Diagnosis ■ Visualization of Radio Waves: Radio Wave Diagnosis ■ Visualization of Wireless Access Points: Traffic Diagnosis *For more details, please download the PDF or feel free to contact us.

• Other services