How to understand COCOEvaluator results: bbox ap50 = seg ap50

[Boatsure]

Training on a custom instance segmentation dataset and all things go well until I find that the AP50 of bbox and seg values are mostly equal at every experiment, while other metrics seem normal. Is that possible and how to understand this phenomenon, please?
In my project, the instance label is an irregular quadrilateral with an approximate rectangular shape and there is only one class. Will this matters?

Thank you all!