Sign in

v1

latestOpenAPI 3.1.02026-07-17114426.0 KB
vision
vision-detect-classify

Detect And Classify

Detect objects then classify each crop — single round-trip.

Runs YOLO detection → crops each bounding box → CLIP classifies each crop. Returns unified results with both detection and classification info.

post/v1/vision/detect_classify

Request body

imagestring required

Base64-encoded image

detection_modelstring

YOLO model for detection

classification_modelstring

CLIP model for classification

classesstring[] required

Classes for zero-shot classification of each crop

confidence_thresholdnumber

Detection confidence threshold

detection_classesstring[] nullable

Filter detections to these YOLO classes

top_kinteger

Top-K classification results per crop

min_crop_pxinteger

Minimum crop dimension in pixels (skip tiny detections)

Response

Successful Response

total_detectionsinteger required
classified_countinteger required
detection_modelstring required
classification_modelstring required
detection_time_msnumber required
classification_time_msnumber required
total_time_msnumber required