We're sunsetting PodQuest on 2025-07-28. Thank you for your support!
Export Podcast Subscriptions
cover of episode Automated cartography using AI
People
D
Daniel Whitenack
G
Gabriel Ortiz
Topics
Daniel Whitenack: 本期节目探讨了人工智能在计算机视觉领域的应用，特别是Gabriel Ortiz在西班牙坎塔布里亚政府的工作，他利用人工智能自动化制图和进行地理空间分析。这展示了人工智能在文本模式之外的巨大潜力。 Gabriel Ortiz: 我拥有30多年的地理空间行业经验，从地形控制、水深测量到工程公司，一直使用地理信息系统(GIS)。我的角色不仅负责数据生产，还负责开发组织内地理空间数据分析的基础设施，我和我的团队将公共服务融入到我们使用人工智能和其他技术的工作中。将深度学习技术与地理空间相结合并非难事，关键在于理解其背后的原理，而非将其视为魔法。我最早接触人工智能是在2012年，当时AlexNet在ImageNet挑战赛中表现出色，但当时图像分类在空间行业应用有限。我早在2010或2011年就关注到英伟达GPU在并行计算方面的潜力，这对于处理栅格数据非常重要。2015年至2016年，一系列深度学习模型架构的出现，例如ResNet和U-Net，对计算机视觉领域产生了重大影响。2018年，我第一次看到Esri公司展示如何使用单次射击检测器自动检测游泳池和石油钻井平台，这让我意识到人工智能的变革潜力。Esri公司正在将许多开源框架集成到其平台中，这简化了深度学习技术在GIS中的应用。我通过线上免费资源学习人工智能基本概念，并采用双管齐下的策略：实践现成模型和学习底层概念。在疫情期间，我们使用深度学习模型计算每个海滩的人数，以应对人群聚集的风险。我们开始使用人工智能模拟领土的各个方面，例如植被检测、城市增长演变和车辆追踪。我们主要使用航空调查数据，但也使用卫星图像数据，并结合激光雷达数据和其他技术。我们使用监督学习和不同的模型架构（如U-Net和Mask R-CNN）进行植被检测。自动化制图是将不同的模型（建筑物、道路、植被等）集成起来，自动生成地图的过程。现代制图利用数字技术，而人工智能则带来了革命性的变化，使地图制作效率更高，质量更好。人工智能可以帮助更新地图数据，但仍需人工干预以纠正错误。人工智能在GIS和制图中的局限性包括计算能力和卷积神经网络（CNN）的固有局限性。为了克服CNN的局限性，我们结合不同的模型架构，并在架构和结果层面进行融合。我对人工智能在GIS领域的未来发展感到兴奋，特别是零样本学习和大型语言模型的结合。 Gabriel Ortiz: 在人工智能辅助下的自动化制图中，我们利用深度学习模型对航空影像进行分析，自动识别并分类地物，例如建筑物、道路、植被等。这极大地提高了制图效率和精度，减少了人工成本和时间。我们还利用人工智能技术对森林扩张、入侵物种和城市增长进行追踪和监测，并对海滩人群进行计数和空间行为分析。这些应用都基于对现有地理信息系统（GIS）工具的扩展和改进，并结合了多种深度学习模型架构，例如ResNet、U-Net和Mask R-CNN。在实际应用中，我们发现模型的准确性受限于计算能力和数据质量，因此我们采用多种策略来提高模型的性能，例如结合不同的模型架构，以及在架构和结果层面进行融合。未来，我们将继续探索零样本学习和大型语言模型等新技术，以进一步提高自动化制图和地理空间分析的效率和精度。

Deep Dive

Key Insights

Why is Spain a good place for AI innovation?

Spain has a strong environment for startups and a growing AI industry, with many professionals in various engineering fields. It encourages engagement with AI talent from the country.

How did Gabriel Ortiz first get interested in AI?

Gabriel's interest in AI began in 2012 with AlexNet, but he also noticed the potential of NVIDIA's GPUs for parallel computing in geospatial data processing as early as 2010.

What challenges does the geospatial industry face with data processing?

Geospatial data, especially raster data like images or digital terrain models, requires significant computational power due to the large number of pixels and complex calculations, such as watershed analysis.

How has Esri integrated AI tools into its platform?

Esri has integrated various open-source AI frameworks into its platform, making it easier for geospatial practitioners to apply AI without needing to manage complex installations or data movement.

What was Gabriel Ortiz's first AI project in GIS?

Gabriel's first major AI project involved using deep learning models to count people on beaches during the COVID-19 pandemic to manage crowd control and provide insights into beach usage patterns.

What types of data does Gabriel Ortiz work with in geospatial analysis?

Gabriel works with aerial surveys, satellite imagery, LiDAR data, and traditional databases. His region, Cantabria, is covered by a national aerial survey plan every three years.

What are some of the limitations of AI in GIS according to Gabriel Ortiz?

Gabriel highlights limitations in computing power and the inherent challenges of convolutional neural networks (CNNs), which may not always perform as expected, even with well-prepared data.

How does Gabriel Ortiz combine different AI models to improve results?

Gabriel combines model architectures, such as using ResNet as a feature extractor with UNET for segmentation, and also combines inference results from different models to overcome individual limitations.

What does Gabriel Ortiz see as the future of AI in GIS?

Gabriel is excited about the potential of Zero Shot learning and the integration of AI models with large language models, which could make geospatial analysis more intuitive and accessible.

Chapters
This segment introduces Gabriel Ortiz, a geospatial expert who transitioned into AI applications. It explores his background in geospatial analysis, his motivations for incorporating AI into his work, and his commitment to public service.
  • Gabriel Ortiz is the principal geospatial information officer at the government of Cantabria, Spain.
  • He has over 30 years of experience in the geospatial industry.
  • His work focuses on applying AI and other technologies to improve public service.
  • He emphasizes the importance of understanding the underlying mechanisms of AI, rather than treating it as a black box.

Shownotes Transcript

Your feed might be dominated by LLMs these days, but there are some amazing things happening in computer vision that you shouldn’t ignore! In this episode, we bring you one of those amazing stories from Gabriel Ortiz, who is working with the government of Cantabria in Spain to automate cartography and apply AI to geospatial analysis. We hear about how AI tooling fits into the GIS workflow, and Gabriel shares some of his recent work (including work that can identify individual people, invasive plant species, building and more from aerial survey data).

Leave us a comment)

Changelog++) members save 3 minutes on this episode because they made the ads disappear. Join today!

Sponsors:

  • Fastly) – Our bandwidth partner. Fastly powers fast, secure, and scalable digital experiences. Move beyond your content delivery network to their powerful edge cloud platform. Learn more at fastly.com)

  • Fly.io) – The home of Changelog.com — Deploy your apps and databases close to your users. In minutes you can run your Ruby, Go, Node, Deno, Python, or Elixir app (and databases!) all over the world. No ops required. Learn more at fly.io/changelog) and check out the speedrun in their docs).

  • Typesense) – Lightning fast, globally distributed Search-as-a-Service that runs in memory. You literally can’t get any faster!

  • Changelog News) – A podcast+newsletter combo that’s brief, entertaining & always on-point. Subscribe today).

Featuring:

Show Notes:

Something missing or broken? PRs welcome!)