In the podcast episode of Intel’s “Code Together” series, AI Solutions for Smart Cities, host Tony Mongkolsmai interviews Markus and Florian of Starwit Technologies to talk about how AI can make our cities smarter using existing infrastructure. They also discuss how AI is changing the development process and the rapidly changing AI landscape.
Innovative Beginnings and Technological Foundations
Founded in April last year by three adept software engineers, Starwit Technologies emerged with a clear mission: to harness the power of open-source software and AI to improve the livability of urban environments. Their focus is on leveraging existing sensoring infrastructure, like cameras, that cities already operate and transforming it into a data source to generate near real-time kinematic models. Those can be used to analyze traffic flows, detect hazards, and aid in urban planning—all without the need for additional infrastructure.
Combining existing open-source algorithms like YOLO to detect and track objects with a deployable and production-ready image processing engine is the foundation of Starwit’s approach to providing data to cities yet unavailable. Starwit’s awareness engine[1] is fully open-source and is already used for several scientific research activities. Thus, Starwit’s solutions are benefitting from any progress being made there.
Embracing Open Source for Collaborative Growth
Open source is thus one of the most compelling aspects of Starwit's strategy. By combining decades of experience in writing and running software with cutting-edge algorithms, Starwit’s awareness engine (SAE) offers a straight path from the latest progress in computer vision-based data generation to deployments in real-world applications. Academic teams can focus on conceptual progress and bypass boilerplate software engineering efforts.
Open source also means that cities around the globe can collaborate and share the cost of software development. Software packages also benefit from a larger user base, and so far, all of Starwit’s products are open source. This move not only accelerates innovation but also aligns with the company's dedication to ensuring data privacy and encouraging community participation in urban development projects.
Challenges of Scaling and Data Management
As with any pioneering venture, Starwit faces its set of challenges, particularly concerning the scalability of its solution and managing vast amounts of data. The co-founders elaborated on the technical hurdles involved in processing data from thousands of cameras and optimizing their system for various hardware setups, from powerful data center equipment to more localized embedded PCs.
Moreover, ensuring the efficient transmission and processing of video data without overwhelming network bandwidth remains a key focus. The goal is to achieve a delicate balance between data fidelity and processing efficiency, enabling the system to provide actionable insights without incurring prohibitive costs. Optimizing algorithms to Intel microprocessors, as well as valuable input from experienced Intel engineers, already boosted the performance of Starwit’s solutions. Benefitting from Intel’s decades of experience in embedded and cloud computing is the main reason for Starwit to join Intel’s Liftoff program. In terms of Intel products, Starwit is looking forward to using Meteor Lake CPUs for performance testing. They continue to use Intel® Extension for PyTorch to optimize runtime performance and are planning to trial the Intel® Geti™ Platform for model training in the second half of 2024.
Towards a Democratic Approach to Urban Management
A central theme of Starwit Technologies' vision is the democratization of urban data. By making kinematic and statistical data accessible to communities, local governments, and businesses, Starwit aims to empower stakeholders to make informed decisions about their urban environments. This approach not only fosters transparency but also encourages a participatory model of urban management, where technology serves as a tool for communal benefit rather than corporate gain.
Looking Ahead
As Starwit Technologies continues to navigate the complexities of smart city innovation, its journey represents a beacon of hope for a future where technology and community converge to create safe, more efficient, and sustainable urban spaces. Their story is a testament to the power of collaborative innovation, open-source development, and a steadfast commitment to improving the quality of urban life for all.
In a world where urbanization continues to pose significant challenges and opportunities, companies like Starwit Technologies stand at the vanguard of transforming our cities into smarter, more responsive environments. Their journey is one to watch, as it encapsulates the promise and potential of smart city technologies to redefine our urban landscapes.
We encourage you to check out Intel’s other AI Tools and Framework optimizations and learn about the unified, open, standards-based oneAPI programming model that forms the foundation of Intel’s AI Software Portfolio.
The Guests:
Markus Zarbock - CEO/CoFounder Starwit Technologies
Florian Stanek - Technical Lead CoFounder Starwit Technologies
References
ChatGPT & Wolf, A. (2023, March 12). Summarize the main points from the following transcript. https://chat.openai.com
[1] https://github.com/starwit/vision-pipeline-k8s
AI Software Marketing Engineer creating insightful content surrounding the cutting edge AI and ML technologies and software tools coming out of Intel
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.