、
代码:
template <typename Dtype> void ROIPoolingLayer<Dtype>::Forward_cpu(const vector<Blob<Dtype>*>& bottom, const vector<Blob<Dtype>*>& top) { //输入有两部分组成,data和rois const Dtype* bottom_data = bottom[0]->cpu_data(); const Dtype* bottom_rois = bottom[1]->cpu_data(); // Number of ROIs int num_rois = bottom[1]->num(); int batch_size = bottom[0]->num(); int top_count = top[0]->count(); Dtype* top_data = top[0]->mutable_cpu_data(); caffe_set(top_count, Dtype(-FLT_MAX), top_data); int* argmax_data = max_idx_.mutable_cpu_data(); caffe_set(top_count, -1, argmax_data); // For each ROI R = [batch_index x1 y1 x2 y2]: max pool over R for (int n = 0; n < num_rois; ++n) { int roi_batch_ind = bottom_rois[0]; //把原图的坐标映射到feature map上面 int roi_start_w = round(bottom_rois[1] * spatial_scale_); int roi_start_h = round(bottom_rois[2] * spatial_scale_); int roi_end_w = round(bottom_rois[3] * spatial_scale_); int roi_end_h = round(bottom_rois[4] * spatial_scale_); //计算每个roi在feature map上面的大小 int roi_height = max(roi_end_h - roi_start_h + 1, 1); int roi_width = max(roi_end_w - roi_start_w + 1, 1); //pooling之后的feature map的一个值对应于pooling之前的feature map上的大小 //注:由于roi的大小不一致,所以每次都需要计算一次 const Dtype bin_size_h = static_cast<Dtype>(roi_height) / static_cast<Dtype>(pooled_height_); const Dtype bin_size_w = static_cast<Dtype>(roi_width) / static_cast<Dtype>(pooled_width_); //找到对应的roi的feature map,如果input data的batch size为1 //那么roi_batch_ind=0 const Dtype* batch_data = bottom_data + bottom[0]->offset(roi_batch_ind); //pooling的过程是针对每一个channel的,所以需要循环遍历 for (int c = 0; c < channels_; ++c) { //计算output的每一个值,所以需要遍历一遍output,然后求出所有值 for (int ph = 0; ph < pooled_height_; ++ph) { for (int pw = 0; pw < pooled_width_; ++pw) { // Compute pooling region for this output unit: // start (included) = floor(ph * roi_height / pooled_height_) // end (excluded) = ceil((ph + 1) * roi_height / pooled_height_) // 计算output上的一点对应于input上面区域的大小[hstart, wstart, hend, wend] int hstart = static_cast<int>(floor(static_cast<Dtype>(ph) * bin_size_h)); int hend = static_cast<int>(ceil(static_cast<Dtype>(ph + 1) * bin_size_h)); int wstart = static_cast<int>(floor(static_cast<Dtype>(pw) * bin_size_w)); int wend = static_cast<int>(ceil(static_cast<Dtype>(pw + 1) * bin_size_w)); //将映射后的区域平动到对应的位置[hstart, wstart, hend, wend] hstart = min(max(hstart + roi_start_h, 0), height_); hend = min(max(hend + roi_start_h, 0), height_); wstart = min(max(wstart + roi_start_w, 0), width_); wend = min(max(wend + roi_start_w, 0), width_); //如果映射后的矩形框不符合 bool is_empty = (hend <= hstart) || (wend <= wstart); //pool_index指的是此时计算的output的值对应于output的位置 const int pool_index = ph * pooled_width_ + pw; //如果矩形不符合,此处output的值设为0,此处的对应于输入区域的最大值为-1 if (is_empty) { top_data[pool_index] = 0; argmax_data[pool_index] = -1; } //遍历output的值对应于input的区域块 for (int h = hstart; h < hend; ++h) { for (int w = wstart; w < wend; ++w) { // 对应于input上的位置 const int index = h * width_ + w; //计算区域块的最大值,保存在output对应的位置上 //同时记录最大值的索引 if (batch_data[index] > top_data[pool_index]) { top_data[pool_index] = batch_data[index]; argmax_data[pool_index] = index; } } } } } // Increment all data pointers by one channel batch_data += bottom[0]->offset(0, 1); top_data += top[0]->offset(0, 1); argmax_data += max_idx_.offset(0, 1); } // Increment ROI data pointer bottom_rois += bottom[1]->offset(1); } }