Natsunoyuki AI Laboratory
Unlock Insights from Space
We specialize in extracting actionable insights from satellite and aerial imagery using computer vision and machine learning. Whether you're in government, agriculture, finance, or logistics, we build custom analytics tools that turn your raw geospatial imaging data into valuable intelligence. Reach out today for an initial conversation!
Industries We Serve
Earth observation data can drive meaningful insights and competitive advantage across a range of sectors. Whether you're optimizing operations, managing environmental risks, or improving forecasting, our analytics solutions are built to serve:1. 🌱 Agriculture: Monitor crop health, estimate yields, and optimize input usage.
2. ⚡ Energy & Utilities: Track infrastructure and assess environmental impact.
3. 🏦 Finance & Insurance: Assess disaster risk, detect fraud, and monitor high-value assets.
4. 🏠Manufacturing & Logistics: Monitor supply chains and infrastructure in remote regions.
5. 🌍 Tourism & Conservation: Monitor land use, environmental change and human impact.
Why Partner with Us?
A Journey Rooted in Purpose
Our story began over seven years ago, driven by a passion for leveraging data to address real world challenges. From national research institutes to startups and leading consulting firms, we've honed our expertise in data analytics and computer vision. Yet, we recognized that our true calling lies in harnessing Earth observation data to create meaningful impact. This realization led us to embark on our entrepreneurial journey with a clear vision: to transform geospatial data into actionable insights that empower industries and communities.Empowering a Sustainable Future
At the heart of our mission is a commitment to sustainability and resilience. We envision a world where every decision, be it in agriculture, infrastructure, or disaster response, is informed by precise, real time Earth observation data. By integrating advanced machine learning techniques, we provide solutions that not only address today's challenges but also anticipate and mitigate future risks.Our Promise to You
We don't just offer services; we offer partnerships. By collaborating closely with you, we tailor solutions that align with your unique objectives and challenges. Together, we can navigate the complexities of our changing world, making informed decisions that lead to sustainable and impactful outcomes.
PeekingDuckReborn: Low Code Computer Vision
PeekingDuckReborn is a modernized evolution of PeekingDuck, an open source, modular Python framework designed for computer vision inference. Originally developed by AI Singapore, PeekingDuck offered a low code approach to building vision pipelines. However, the original project is no longer maintained and has become incompatible with modern Python libraries and tooling.We created PeekingDuckReborn to revitalize this powerful idea, bringing it up to date with modern libraries, improved performance, and long term maintainability.
Why Deployment Matters
Training and developing machine learning models is only part of the journey. Equally important is deploying those models in ways that clients can use reliably, efficiently, and at scale. This includes model packaging, inference optimization, visualization, and post deployment maintenance. These challenges often require significant engineering effort beyond model training itself.
A Low Code Modular Framework for Vision Inference
PeekingDuckReborn was born out of a need to solve these challenges. It provides a modular, low-code framework for deploying customizable computer vision pipelines without writing extensive inference or visualization code. With built in support for:
1. Object detection and tracking,
2. Pose estimation,
3. Zone-based logic and analysis,
4. Live video or image input/output,
5. Real-time overlay visualization,PeekingDuckReborn is designed to help teams move faster from prototype to production by simplifying the entire deployment process.
Rapid Pipeline Creation with YAML
For example, a pipeline to detect, track, and visualize the positions of cars on a busy street can be assembled with just a few lines of YAML configuration:
Different functionalities such as object detection, object tracking, and zone-based crowd analysis can be easily mixed and matched to support a wide range of use cases.
Visual Example: Real Time Traffic Monitoring
This inference result was created without writing any inference logic or drawing code, just declarative YAML configuration. PeekingDuckReborn handles the rest.
Open Source and Actively Maintained
PeekingDuckReborn is fully open source and freely available on GitHub. It is actively maintained with support for modern Python libraries, improved tracking algorithms, clearer visual outputs, and better runtime compatibility.⚠️ Note: The original PeekingDuck project by AI Singapore is no longer actively maintained and may not work with modern systems. PeekingDuckReborn is a modern, independent fork designed to carry the project forward.
How We Work and Our Fees
Our R&D Process
Problem Discovery: We begin by thoroughly understanding your challenges, formulating precise problem statements, defining clear objectives, and validating potential solution approaches.
Proof-of-Concept: We develop and test a prototype to assess feasibility, refine methodologies, and establish a foundation for scalable, effective solutions.
Product Enhancement: We focus on optimizing your product’s performance and impact, unlocking greater functionality, and delivering more value to your users.
Our Fees
We believe in aligning our fees with the value we deliver to ensure a fair and mutually beneficial partnership. Instead of traditional hourly billing, we adopt a value-based pricing model. Typically, our fees are a percentage of the return on investment generated through our contributions. This approach ensures that you receive the best value from our services.
About Us
We are a R&D lab dedicated to shaping the future of Earth observation through machine learning and satellite image analysis. Drawing on years of deep expertise in computer vision, machine learning, and analytics consulting, we are at the forefront of this dynamic and rapidly evolving field.Our team has a proven track record of solving complex visual problems across diverse industries. Now, we’re channeling that expertise to unlock actionable insights from satellite imagery, creating new opportunities for sectors such as environmental monitoring, urban development, agriculture, and beyond.Driven by a culture of agility, curiosity, and practical innovation, we move fast and build solutions that deliver real world impact. We seek to partner with forward thinking organizations eager to harness the potential of Earth observation data, whether you’re well versed in the field or just beginning your exploration.Let’s discover what we can achieve together.
Example Use Cases
Earthquake Damage Monitoring
Leverage satellite based computer vision to rapidly assess environmental, structural, and infrastructure damage following major seismic events. The animations above illustrate ground deformation detected after recent earthquakes.Left: Mandalay International Airport region during the 2025 March 28 Myanmar earthquake. The imagery reveals vertical ground displacement near the airport, with uplift occurring on one side of the main road.Right: Kaiso Port area during the 2024 January 1 Noto Peninsula earthquake. Here, significant coastal movement is visible, as the shoreline shifts inland by approximately 250 meters.Such rapid, large scale assessments are critical for emergency response planning, infrastructure triage, and long term recovery strategies.
Landslide and Avalanche Monitoring
Use satellite imagery and computer vision to rapidly estimate the geographic extent of ground disturbance from natural hazards such as landslides, avalanches, or glacier collapses. This enables more accurate and timely estimation of potential casualties, economic losses, and infrastructure damage.The example above shows the aftermath of the 2025 May 28 glacier collapse near Blatten, Switzerland. By analyzing high resolution satellite data, we estimate the affected area to be approximately 2.26 km², providing critical spatial context for emergency response and recovery efforts.
Airport Activity Monitoring
Track indirect economic indicators from space, such as airport and port activity to support forecasting, risk assessment, and market intelligence. The imagery above demonstrates how our custom aircraft detection model automatically identifies aircraft in very high resolution satellite imagery, enabling continuous monitoring of airport activity.This data can reveal patterns in passenger and cargo traffic, highlight changes in regional mobility, and serve as a proxy for broader economic conditions, all without the need for on the ground access.
Precision Agriculture
Enhance agricultural decision making with satellite and aerial imagery powered by computer vision. By detecting and monitoring crop growth at scale, stakeholders can better estimate harvest yields, optimize inputs, and mitigate agricultural risks.In the example above, our custom detection model identifies and counts individual cabbage plants from high resolution aerial imagery. This level of crop specific insight enables more accurate forecasting, improved resource allocation, and smarter risk hedging across the agricultural value chain.
Contact Us
Please submit your queries using the form below. A representative will follow up with you shortly.