Skip to main content

roboflow-supervision

Usa a post-procesar detecciones y anotar con Supervision.

Jump to install

Source facts

Repository
Ntizar/MasterMind
Last source activity
September 5, 2026 at 21:34
Detected SKILL.md language
Spanish
Stars
2
Forks
0

Install options

The review-first prompt is selected by default. You can switch to a direct command or download a local copy.

Review the source files

Read SKILL.md and any companion files shown by SkillsMP before deciding whether to install.

Showing SKILL.md

SKILL.md
Source instructions · Read-only preview
name
roboflow-supervision
description
Usa a post-procesar detecciones y anotar con Supervision.
version
2.0.0
tags
["supervision","deteccion","anotar","yolo","roboflow","vision","python"]
related_skills
["roboflow-supervision","cctv-yolo","fast-alpr","rf-detr"]
# Supervision — post-procesado y anotaciones de visión (API actual) > ⚠️ Corrección 2026-09-05 (auditoría): `sv.Image`, `Detections.from_yolov8` y `Detections.from_coco_json` ya **no existen** en Supervision moderno. La API actual pasa numpy/PIL directo y usa `from_ultralytics` y helpers de `supervision.dataset`. **Repo:** `https://github.com/roboflow/supervision` (Python, ~50K⭐). ## When to Use - Cuando pidas **anotar/visualizar/dibujar cajas** de detecciones (YOLO, etc.) o filtros/post-procesado (zone, line, byte_tracker) sobre vídeo/imagen. ## Uso (API real) ```python import supervision as sv # Anotar con cajas: la imagen se pasa directa (numpy/PIL), sin sv.Image annotated = box_annotator.annotate(scene=image.copy(), detections=detections) # Detecciones desde Ultralytics YOLO: detections = sv.Detections.from_ultralytics(results) # (ya no from_yolov8) # Carga de datasets COCO: helpers de supervision.dataset from supervision.dataset import load_coco_annotations, coco_annotations_to_detections ``` ## Pitfalls - **No** `sv.Image` — pasa la imagen numpy/PIL directamente a `annotate(scene=..., detections=...)`. - **No** `Detections.from_yolov8` → `from_ultralytics`; `from_yolov5` sí existe. - **No** `Detections.from_coco_json` → helpers `supervision.dataset`. ## Verificación - `annotate(scene=image, detections=...)` y comprobar que las bboxes se dibujan; `from_ultralytics` con un modelo YOLO.
View on GitHub