Some questions about input and output.

[A-good-kids]

First of all, thank you for your outstanding contributions. I have some confusion when introducing your method into my own project. The size of my input image is 550550, the size of the feature map is 69, 69, and the roi is tensor [1, 12, 20 , 331, 415] roi_features = prroi_pool2d(featuremap,roi, 7,7, 1/8) Is it correct to use this way? Then output the result roi_shape: torch.Size([5, 256, 7, 7]) , 2567*7 size feature, and what does this 5 mean? If you can get your answer, it will be of great help to me.

[vacancy]

ROI should be a 2-dim tensor.

Feature Map = [B, C, H, W] ROI = [K, 5] Output = [K, C, 7, 7]

Do not duplicate your question.

[A-good-kids]

What does k mean?

[vacancy]

Number of ROIs.