Technology
Detectron2
Meta AI's high-performance engine for state-of-the-art object detection and image segmentation built on PyTorch.
Detectron2 provides a modular framework for computer vision tasks: object detection, instance segmentation, and panoptic segmentation. Built by Meta AI on the PyTorch backend, it powers production systems and research benchmarks (including Mask R-CNN and DensePose). The library features a comprehensive Model Zoo with pre-trained weights for datasets like COCO and Cityscapes. Developers use its extensible API to swap backbones or custom heads, ensuring rapid iteration from research code to large-scale deployment.
Related technologies
Recent Talks & Demos
Showing 1-1 of 1