Manual sperm motility assessment using microscopy is challenging. It requires extensive training, making computer-assisted sperm analysis (CASA) a popular alternative. However, supervised machine learning needs more data to enhance accuracy and reliability. To address this, researchers have introduced the VISEM-Tracking dataset, containing 20 video recordings (29,196 frames) of wet sperm preparations with manually annotated bounding-box coordinates and expert-analyzed sperm characteristics.
The VISEM-Tracking dataset stands out as a unique and invaluable resource for advancing computer-assisted sperm analysis by providing a comprehensive collection of diverse, high-quality sperm video recordings with expert annotations. This addresses the current lack of data and facilitates the development of more accurate and reliable machine-learning models in sperm motility assessment.
The YOLOv5 deep learning model, trained on the VISEM-Tracking dataset, has shown promising baseline sperm detection performance, indicating the dataset's potential for training complex models to analyze spermatozoa.
VISEM-Tracking is available on Zenodo under a Creative Commons Attribution 4.0 International (CC BY 4.0) license, featuring 30-second videos from 20 different patients with annotated bounding boxes. Additional 30-second video clips from both the annotated and unlabelled portions of the VISEM dataset are also provided, making it suitable for future research in semi- or self-supervised learning.
Access the dataset here: https://zenodo.org/record/7293726
Read the full paper: https://www.nature.com/articles/s41597-023-02173-4
Please sign in or register for FREE
If you are a registered user on Research Data at Springer Nature, please sign in