For an overview of Flukebook's multi-species, multi-modal machine learning
for photo ID, please see:https://docs.wildme.org/docs/researchers/ia_pipeline
Community support for Wildbook is provided at:
NEW DEVELOPMENTS
- SAGE-351: New, machine learning-based sperm whale fluke ID algorithm added using PIE V2 (triplet loss network).
- New sperm whale fluke detector and related detection pipeline models, including orientation network support.
- Data integrity checks now include a check for viewpoint = null on annotations
- Bulk import now handles and displays progress of acmID registration of uploaded images
- Bulk import review page improvements have been made.
- Social relationships and co-occurrence diagrams and tables now load faster.
- WB-1785: Project matching has improved user feedback regarding IA status.
- Established a default setting for new Wildbook users to not receive email alerts.
- Measurements now available for standard export.
- Collaborations expanded to include public data for non-silo communities.
- Org admins can add managing researchers to encounters.
BUG FIXES
- WB-1854: Encounter Search mark-recapture search includes capture sessions that occur on the same day as the session end time.
- WB-1808: The “Add to existing individual ID” function no longer has an unneeded dependency on the ShowTaxonomy configuration option.
- SAGE-391: Background subtraction failing with new model files exported with NumPy float64 data type
- SAGE-386: Background subtraction added missing support for bottlenose_dolphin+fin_dorsal
- WB-1781: User can now self-edit email addresses.
- WB-1740: Dates on images in match results now reflect user-provided dates rather than exif.
- WB-1479: Scroll wheel actions work with search results tables.
