We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations

DBLP - CS Bibliography


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Computer Vision and Pattern Recognition

Title: CytoImageNet: A large-scale pretraining dataset for bioimage transfer learning

Abstract: Motivation: In recent years, image-based biological assays have steadily become high-throughput, sparking a need for fast automated methods to extract biologically-meaningful information from hundreds of thousands of images. Taking inspiration from the success of ImageNet, we curate CytoImageNet, a large-scale dataset of openly-sourced and weakly-labeled microscopy images (890K images, 894 classes). Pretraining on CytoImageNet yields features that are competitive to ImageNet features on downstream microscopy classification tasks. We show evidence that CytoImageNet features capture information not available in ImageNet-trained features. The dataset is made available at this https URL
Comments: Accepted paper at NeurIPS 2021 Learning Meaningful Representations for Life (LMRL) Workshop
Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Quantitative Methods (q-bio.QM)
Cite as: arXiv:2111.11646 [cs.CV]
  (or arXiv:2111.11646v2 [cs.CV] for this version)

Submission history

From: Stanley Bryan Hua [view email]
[v1] Tue, 23 Nov 2021 04:37:15 GMT (259kb,D)
[v2] Wed, 24 Nov 2021 04:26:05 GMT (259kb,D)

Link back to: arXiv, form interface, contact.