The region of interest (RoI) pooling layer can be implemented in deep learning frameworks like PyTorch using `torchvision.ops.roi_pool`. The function takes the CNN-extracted feature map, a tensor of region proposals, the desired output size, and a spatial scale factor. Each region proposal is represented by five elements: the object class index, followed by the $$(x, y)$$-coordinates of its upper-left and lower-right corners. Because the feature map is typically downsampled relative to the original image, the `spatial_scale` argument is used to scale the region proposal coordinates from the original image space to the feature map space before pooling.

```python
import torch
import torchvision

X = torch.arange(16.).reshape(1, 1, 4, 4)
rois = torch.Tensor([[0, 0, 0, 20, 20], [0, 0, 10, 30, 30]])
torchvision.ops.roi_pool(X, rois, output_size=(2, 2), spatial_scale=0.1)
```

Region of Interest Pooling Code Implementation

Consider an illustrative example of a region of interest (RoI) pooling layer computation. Suppose we select the upper-left $$3 	imes 3$$ region of interest on a $$4 	imes 4$$ input matrix and apply a $$2 	imes 2$$ RoI pooling layer. This pooling layer divides the $$3 	imes 3$$ region into a $$2 	imes 2$$ grid of subwindows to obtain a $$2 	imes 2$$ output. Because the subwindow dimensions must be integers, they are rounded up. The four resulting subwindows contain the following elements from the input matrix: $$0, 1, 4,$$ and $$5$$; $$2$$ and $$6$$; $$8$$ and $$9$$; and $$10$$. By extracting the maximum value from each of these subwindows, the RoI pooling layer produces an output matrix containing $$5, 6, 9,$$ and $$10$$.

Region of Interest Pooling Computation Example

The region of interest (RoI) pooling layer is introduced in Fast R-CNN to extract fixed-shape features from regions of interest of varying sizes. Unlike standard pooling layers that indirectly control output shape through window size, padding, and stride, RoI pooling allows the output shape (e.g., $$h_2 \times w_2$$) to be specified directly. For an input region of shape $$h \times w$$, the RoI pooling layer divides the region into an $$h_2 \times w_2$$ grid of subwindows, where each subwindow is approximately of shape $$(h/h_2) \times (w/w_2)$$. The subwindow dimensions are rounded up, and the maximum element in each subwindow is taken as its output.

Claude

The Fast R-CNN model processes an image through four major computational steps. First, a trainable CNN extracts features from the entire image, outputting a feature map of shape $$1 	imes c 	imes h_1 	imes w_1$$. Second, selective search generates $$n$$ region proposals, which are mapped as regions of interest on the CNN output. A region of interest (RoI) pooling layer then extracts concatenated features of a uniform shape $$n 	imes c 	imes h_2 	imes w_2$$ from these proposals. Third, a fully connected layer transforms these features into a matrix of shape $$n 	imes d$$. Finally, the output is transformed into a shape of $$n 	imes q$$ for object class prediction using softmax regression (where $$q$$ is the number of classes) and a shape of $$n 	imes 4$$ for bounding box prediction.

Learn Before

Related

Learn After