FlyPix AI Unveils Open-Access Trash Detection Model on Earth Day 2025

In celebration of Earth Day 2025, FlyPix AI has officially launched its AI-powered trash detection model, making it publicly available to support environmental monitoring and clean-up efforts worldwide. This release represents a significant advancement in the application of artificial intelligence and drone technology for environmental protection and sustainability.

Developed to process high-resolution drone imagery, the model can detect and classify over 20 distinct categories of waste, including plastic packaging, tires, cans, and bottles. In addition to object classification, the system delivers real-time analytics on litter density and surface coverage, enabling efficient site assessments and informed decision-making for environmental teams.

The model’s capabilities are designed to address a growing need for scalable solutions in waste tracking and habitat restoration. By providing accurate, automated assessments of polluted areas, it reduces the need for manual surveys and helps organizations direct resources to areas with the highest impact potential.

an environmental NGO working to remove plastic waste from coastlines. Through this partnership, Second Life is integrating FlyPix AI’s detection model into its drone operations to improve shoreline surveillance, trash quantification, and the overall effectiveness of its clean-up missions.

To encourage transparency and collaboration, registered users can explore the performance of the model and its real-world outputs in the Community Map section of the FlyPix AI platform. This interactive feature showcases publicly shared results from various regions, allowing users to evaluate detection quality and contribute their own data to the growing global map of environmental waste.

The trash detection model is available via the FlyPix AI platform at https://flypix.ai.

FlyPix AI remains committed to leveraging geospatial AI to support a cleaner, more resilient planet — combining technology, accessibility, and real-world impact.