Detectron2
A Python computer vision library for object detection and segmentation, with pretrained models and deployment exports under Apache 2.0.
Detectron2 is a Python library for developers and researchers building object detection and image segmentation applications. It provides model training tools and reusable components for computer vision research and production software.
- Detection and segmentation: Work with object detection, panoptic segmentation, rotated bounding boxes and Densepose, with approaches including Cascade R-CNN, PointRend and DeepLab.
- Pretrained models: Use downloadable Model Zoo models as starting points and compare experiments against published baseline results.
- Research development: Build projects on the library using supported approaches such as ViTDet and MViTv2.
- Deployment exports: Export models to TorchScript or Caffe2 for use outside the training workflow.
Facebook AI Research developed Detectron2 as the successor to Detectron and maskrcnn-benchmark. It's open source under the Apache 2.0 license.
Logotyper
