Repository of community practices in Ocean Research, Applications and Data/Information Management
  • About
  • FAQs
  • About
  • FAQs
    • Login
    View Item 
    •   Repository OceanBestPractices
    • ⇒ ZZZ_Miscellaneous Communities
    • Miscellaneous Community Practices
    • View Item
    •   Repository OceanBestPractices
    • ⇒ ZZZ_Miscellaneous Communities
    • Miscellaneous Community Practices
    • View Item
    JavaScript is disabled for your browser. Some features of this site may not work without it.

    Best Practices for Fine-tuning Visual Classifiers to New Domains.

    Thumbnail
    View/Open
    PDF (Pre-print) (139.2Kb)
    Average rating    votes
    Date
    2016
    Author
    Chu, Brian
    Madhavan, Vashisht
    Beijbom, Oscar
    Hoffman, Judy
    Darrell, Trevor
    Status
    Published
    
    Metadata
    Show full item record
    Abstract
    Recent studies have shown that features from deep convolutional neural networks learned using large labeled datasets, like ImageNet, provide effective representations for a variety of visual recognition tasks. They achieve strong performance as generic features and are even more effective when fine-tuned to target datasets. However, details of the fine-tuning procedure across datasets and with different amount of labeled data are not well-studied and choosing the best fine-tuning method is often left to trial and error. In this work we systematically explore the design-space for fine-tuning and give recommendations based on two key characteristics of the target dataset: visual distance from source dataset and the amount of available training data. Through a comprehensive experimental analysis, we conclude, with a few exceptions, that it is best to copy as many layers of a pre-trained network as possible, and then adjust the level of fine-tuning based on the visual distance fro.....
    Resource URL
    http://adas.cvc.uab.es/task-cv2016/papers/0002.pdf
    Title of Book
    Computer Vision – ECCV 2016 Workshops: Amsterdam, The Netherlands ..., Part 3.
    Editor(s) of Book
    Hua, Gang
    Jégou, Hervé
    Page Range
    pp.435-442
    Publisher
    Springer
    Switzerland
    Document Language
    en
    Sustainable Development Goals (SDG)
    14.A
    Essential Ocean Variables (EOV)
    Zooplankton biomass and diversity
    Maturity Level
    TRL 4 Component/subsystem validation in laboratory environment
    Best Practice Type
    Best Practice
    Citation
    Chu, B. et al (2016) Best Practices for Fine-tuning Visual Classifiers to New Domains. In: Comouter Visions, ECCV 2016 Workshop. (eds. G. Hua and H. Jégou). Switzerlnd, Springer International Publishings, 8pp. DOI: http://dx.doi.org/10.25607/OBP-765
    URI
    http://hdl.handle.net/11329/1250
    http://dx.doi.org/10.25607/OBP-765
    Collections
    • Miscellaneous Community Practices [469]

    What results are displayed?

    Perform Semantic Advanced Search.

    Browse

    All of OceanBestPracticesCommunities & CollectionsBy Issue DateAuthorsTitlesSubjectsThis CollectionBy Issue DateAuthorsTitlesSubjects

    My Account

    LoginRegister

    Statistics

    View Usage StatisticsView Google Analytics Statistics

    © . UNESCO/IOC Project office for IODE Oostende, Belgium.
    Help Desk | Send Feedback

    How to Cite: OceanBestPractices Repository. Oostende, Belgium, International Oceanographic Data and Information Exchange(IODE). [Database]. Available: https://repository.oceanbestpractices.org/.

            

    Disclaimer: IOC/IODE does not warrant that the information, documents and materials contained in the OceanBestPractices repository website is complete and correct and shall not be liable whatsoever for any damages incurred as a result of its use.
    The information, data, statements, geographical boundaries, maps and declarations expressed in documents in this repository do not imply the official endorsement or acceptance by UNESCO. Contributors to this repository are solely responsible for the contents of their uploaded documents, the inclusion of which in this repository does not mean the approval or acceptance by UNESCO.

    Privacy Policy