OD
Object detection
Build bounding box detection pipelines for images and video frames using configurable models.
Detectron2 is designed for advanced detection and segmentation workflows while staying flexible for custom research.
Build bounding box detection pipelines for images and video frames using configurable models.
Detect object-level masks and handle complex scenes where object boundaries matter.
Combine things and stuff classes for richer scene understanding and visual analysis.
Use pose and landmark-style workflows where point-level predictions are required.
Register COCO-style or custom datasets and tune training for your own data.
Prepare models for production and integration with larger application pipelines.