Organisation Name

AI Lab at German Environment Agency

Industry

Public service

Organisation Website

https://www.umweltbundesamt.de/en/topics/digitalisation/the-ai-lab-at-the-german-environment-agency

Country

Germany

Sustainable Development Goals (SDGs)

General Description of the AI tool

This AI-powered tool combines image classification and rule-based text analysis to detect illegal online wildlife trade. Utilizing a tailored web scraper, energy-efficient database, and user-friendly frontend, it processes large-scale data flows to identify protected species. With specialized expert knowledge and automation, it overcomes the challenge of manually differentiating over 40,000 species. Developed as a prototype by the AI Lab for Germany's Federal Agency for Nature Conservation, this tool has the potential to support monitoring of illegal species trade.

Relevant Research and Publications

GroundingDINO (open-source AI framework) – Provided advanced capabilities for visual grounding, enabling our system to crop original images to relevant animals.

YOLO (You Only Look Once, open-source AI object detection model) – Served as the backbone for real-time object detection, allowing the system to automatically classify thousands of animal images with high accuracy.

Kulkarni, 2022: “Towards automatic detection of wildlife trade using machine vision models” – Offered methodological inspiration and demonstrated the feasibility of applying machine vision to wildlife trade detection, validating our interdisciplinary approach.

iNaturalist Dataset (https://github.com/visipedia/inat_comp/tree/master/2021) Provided a large, high-quality dataset of labeled species images that was essential for training and validating our AI models, ensuring coverage across many CITES-listed species

Needs