search

mkang315 / ASF-YOLO

[IMAVIS] Official implementation of "ASF-YOLO: A Novel YOLO Model with Attentional Scale Sequence Fusion for Cell Instance Segmentation".
GNU Affero General Public License v3.0
63 stars 3 forks source link

有一个论文问题想请教一下作者 #2

Open Ellohiye opened 8 months ago

mkang315 commented 8 months ago

"In SSFF, The P3, P4, and P5 feature maps are normalized to the same size, upsampled, and then stacked together as input to a 3D convolution to combine multiscale features." See Page 6.

mkang315 commented 6 months ago

Thanks for pointing this problem out. To clarify the expression, we replace scale-invariant features' withaspect ratio invariant features'. As you know, each learned filter of Convolutional neural networks (CNNs), including upsampling and downsampling, is sensitive to a given set of features only within a narrow range of scale. During the process, all images are resized, but the aspect ratio, which is the ratio of their width to height, is preserved.