search

ing-oasis / P2PNet-with-backbone-from-YOLOV6

This model replaced Vgg16-bn and decoder in P2PNet with RepVGG and Rep-PAN. However, the number of channels keeps decreasing in Rep-PAN, which may pose a negative effect when fusing different feature maps. And it doesn't perform as good as expected.
Apache License 2.0
0 stars 0 forks source link

some difference #1

Open flywheel1412 opened 1 year ago

flywheel1412 commented 1 year ago

i think the official network's head networks using P3 feature map, and you's using P4 feature map, this maby degrade the precision

ing-oasis commented 1 year ago

You got the point. I will give it a try when I get my server back:)

flywheel1412 commented 1 year ago

i found that many crowd counting network using VGG16 as the backbone to extract the feature, maybe this task rely on the low level feature and the deeply and powerful backbone will not work well (i tried some large backbones), maybe shallow and wide backbone networks and powerful FPN, i'm trying for this hypothesis, hope it's helpful for u.