search

facebookresearch / detectron2

Detectron2 is a platform for object detection, segmentation and other visual recognition tasks.
https://detectron2.readthedocs.io/en/latest/
Apache License 2.0
30k stars 7.41k forks source link

For Panormic FPN, how do I output the target detection box when I'm using the trained model to test the image/video? #596

Closed tensorflowt closed 4 years ago

tensorflowt commented 4 years ago

hi First of all, thank you very much for your open source project, very excellent! Here is a question to ask you!

For Panormic FPN, how do I output the target detection box when I'm using the trained model to test the image/video? At present, I can only output instance segmentation and panorama segmentation information.

What changes do I need to make during the test? Or does the algorithm not support the output of the det box?

ppwwyyxx commented 4 years ago

In the demo we only draw Panoptic Segmentation, but the output of Panoptic FPN contains the "instances" key and therefore contains bounding boxes. Model output format is described in https://detectron2.readthedocs.io/tutorials/models.html#model-output-format