Computer Vision for Natural Heritage

---

Large-scale digitization initiatives are generating massive image datasets from natural history collections. Computer vision is essential for unlocking the scientific value of these data, enabling automated extraction of specimen information and supporting research in biodiversity, ecology, and evolution, including studies of migration and ongoing mass extinction. However, despite controlled imaging conditions and rich metadata, automated analysis remains challenging due to complex specimen structures, varying appearances, and handwritten labels.

The Computer Vision for Natural Heritage (CVNH) is an ECCV workshop that brings together computer vision researchers, natural heritage digitization experts, and domain scientists to advance methods for analyzing 2D, 3D, and multi-modal imaging of natural history collections and to identify open challenges.

CVNH workshop has a Call for Papers and presents two Challenges:

• Foram2026: Detection and classification of microCT 3D scans of Forameniferas,
• MuseumLabels2026: Optical Character Recognition of Museum label images.

The top-performing teams of the challenges will be invited to participate in a paper related to the challenge.