Sourcing field guide
Build an image classification taxonomy that survives review
Define visual classes, uncertain cases, and label relationships before requesting a large photo collection.
← Photo Datasets
Fiund / Resources
The resource library.
Original guides to photo datasets: source selection, collection design, annotations, evaluation, and project review.
27 guides shown
Sourcing field guide
Specify bounding boxes for an object detection dataset
Set practical boundaries for object instances, truncation, occlusion, and box review.
Sourcing field guide
Choose segmentation masks that match the visual task
Compare semantic regions and individual instances while specifying edges and difficult materials.
Sourcing field guide
Plan product photographs beyond the catalog hero shot
Specify product identity, variants, packaging, and real-use imagery for visual models.
Sourcing field guide
Scope indoor and outdoor scene image collections
Build a useful scene collection through activity context, layout, and environmental variation.
Sourcing field guide
Find near-duplicate photographs before counting scale
Distinguish unique visual coverage from bursts, crops, exports, and repeated source material.
Sourcing field guide
Design photo dataset splits around the real deployment question
Choose grouping rules that test new subjects, locations, products, or capture conditions.
Sourcing field guide
Buy useful class coverage instead of equal image counts
Set coverage targets that distinguish common classes, scarce examples, and review priorities.
Sourcing field guide
Specify negative examples for visual recognition
Define what a model should reject and how to distinguish negatives from unlabeled positives.
Sourcing field guide
Collect hard visual examples without losing representativeness
Turn observed errors into a targeted photo acquisition brief with reviewable coverage.
Sourcing field guide
Describe geographic coverage without overclaiming representation
Specify locations, environmental features, and reporting limits for a photo sourcing project.
Sourcing field guide
Specify lighting variation for computer vision photographs
Make exposure, shadows, and illumination conditions concrete in a photo collection brief.
Sourcing field guide
Plan viewpoint coverage for visual recognition
Specify angle, distance, framing, and orientation instead of requesting many similar photos.
Sourcing field guide
Write a clear occlusion policy for image datasets
Define partial visibility, uncertain identity, and annotation behavior for hidden objects.
Sourcing field guide
Set image resolution requirements around visible detail
Translate model needs into useful source and delivery image specifications.
Sourcing field guide
Review image compression and visual artifacts
Assess delivered photographs for artifacts that affect the intended vision task.
Sourcing field guide
Keep original and derived photo assets traceable
Document crops, edits, transformations, and source relationships in an image delivery.
Sourcing field guide
Build a photo dataset manifest that answers buyer questions
Choose identifiers, metadata fields, and review states that make an image delivery usable.
Sourcing field guide
Specify image and text pairs for visual language work
Define grounded captions, descriptions, and metadata without treating all text as equivalent.
Sourcing field guide
Source text-bearing images for OCR evaluation
Describe legibility, layout, language, and transcription rules for text visible in photographs.
Sourcing field guide
Create an acceptance plan for photo annotations
Turn annotation review into a reproducible process with error categories and correction rules.
Sourcing field guide
Version photo datasets without confusing downstream users
Track additions, removals, corrections, and split changes across image dataset releases.
Sourcing field guide
Review a photo dataset sample before scaling acquisition
Use representative images and metadata to test whether a sourcing proposal fits the actual task.
Sourcing field guide
Understand the scope choices behind photo sourcing costs
Compare acquisition proposals through source diversity, annotations, processing, and review effort.
Sourcing field guide
Choose an existing photo archive or a new collection
Compare speed, control, source documentation, and coverage gaps without assuming either path fits.
Sourcing field guide
Scope a photo dataset for computer vision
Define classes, capture variation, annotation, and source requirements for image sourcing.
Sourcing field guide
Review image annotation quality before scaling
Use clear label rules and a representative sample to evaluate boxes, masks, and classes.