shivangi-aneja / COSMOS

[AAAI 2023] COSMOS: Catching Out-of-Context Misinformation using Self Supervised Learning
MIT License
41 stars 14 forks source link

How did you choose maskrcnn_bboxes? #7

Open kondounagi opened 2 years ago

kondounagi commented 2 years ago

Usually, Mask-R-CNN outputs many bounding boxes to an image. In your paper, the model use only ten bounding boxes for an image. How did you choose ten bounding boxes? (confidence, the size of bbox, ...)

maskrcnn_bboxes List of detected bounding boxes corresponding to the image. (x1,y1) refers to start vertex of the rectangle and (x2, y2) refers to end vertex of the rectangle

Thanks