Sign inGet started
Dataset · Hugging Face

FIVE-MGI/SiDoLa-NS-Micro-PNS

SiDoLa-NS-Micro-PNS Dataset Summary This dataset contains high-resolution microscopy images of peripherla nervous system (PNS) microstructures, together with their corresponding labels for training segmentation and detection models. The dataset is primarily intended for training and benchmarking deep learning pipelines (e.g., YOLO, SAHI, SAM-based workflows). In addition to the raw images and labels, some dataset folders also contain: Pretrained PyTorch models… See the full description on the dataset page: https://huggingface.co/datasets/FIVE-MGI/SiDoLa-NS-Micro-PNS.

Author / organizationFIVE-MGI
License (as declared)apache-2.0
Downloads182 (as reported by Hugging Face)
Likes0
Created26 December 2024
Last updated at source30 September 2025
microscopysegmentationyolopytorchtrained-modelsperipheral nervous systemaxonsmyelin sheaths
Embed this entry
<iframe src="https://silytics.si/embed/item/59ec51f2-f9cc-4b87-bba3-c00d40827e96" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Free, no tracking. More about feeds and embeds.

Popularity, last 3 days

182 now. Counts as reported by the source, snapshotted daily by Silytics.

Recent changes

  • 16 Oct 2026 Description: SiDoLa-NS-Micro-PNS Dataset Summary This dataset contains high-resolution microscopy images of peripherla nervous system (PNS) microstructures, together with their corresponding labels for training segmentation and detection models. The dat → SiDoLa-NS-Micro-PNS Dataset Summary This dataset contains high-resolution microscopy images of peripherla nervous system (PNS) microstructures, together with their corresponding labels for training segmentation and detection models. The dat

Changes detected by Silytics between refreshes of the source.

More from FIVE-MGI

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 18 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.