Datasets

CVNH workshop introduces two dataset challenges: Foram2026 and MuseumLabel2026.

Submissions

Format and platform: The predictions will be submitted as a CSV file in Kaggle. The exact format (expected rows and columns) is specified in the corresponding Kaggle challenge page.

Deadline: August 31, 2026.

Winners: The winners of the challenges will be announced at the workshop. The winners will receive a signed diploma, and they will be invited to participate in a paper related to the datasets.

Short description: We ask all participants to submit a brief description of their method (max. 1 page PDF) before the deadline. A link to submit this description will be provided. Although this is not a requirement for participating in the challenge, it is a requirement to qualify as a winner and being invited to participate in the paper.

One slide: Similarly to the “Short description”, we ask all participants to provide one slide explaining their approach. All slides will be combined and shown at the workshop. Although this is not a requirement for participating in the challenge, it is a requirement to qualify as a winner and being invited to participate in the paper.

Foram2026 Challenge - Kaggle Competition

Task: Detection and classification of microCT 3D scans of Forameniferas.

Dataset size: 2425 labelled 3D volumes of individual forams + 95 unlabelled 3D volumes contained mixed specimens.

Example volume:

Training set

Test set

MuseumSCAT2026 Challenge - Kaggle Competition

Task: Text recognition and text type identification (e.g., “date”, “locality”) in museum label photographs.

Dataset size: 3500 images and annotations.

Example image:

Sign up in our newsletter to be notified about updates.