Detectron2 is the object detection open source project based on the pytorch made in the Facebook AI Research (FAIR). With modular design, Detectron2 is more flexible, extensible than the existing Detectron. Detectron2 provides models of object detection such as panoptic segmentation, DensePose, Cascade RCNN, and more based on a variety of backbones.
In this Ainize project, you can receive the inferred result image after selecting one of the inference models. All the inference models used Resnet 50 + FPN (Feature Pyramid Network) as a backbone.
The inference using server is done in the following steps: