In the pointed head module, did you use the corner pooling?

microsoft / RelationNet2

RelationNet++: Bridging Visual Representations for Object Detection via Transformer Decoder

MIT License

209 stars 27 forks source link

In the pointed head module, did you use the corner pooling? #3

Open b03505036 opened 3 years ago

b03505036 commented 3 years ago

In the pointed head module, did you use the corner pooling, like the CornerNet way? or directly using feature maps of each level to predict the corner's heatmap of both tl and br ?

lxtGH commented 3 years ago

The same question.

b03505036 commented 3 years ago

Hope I can get the reply . I sent the e-mail for weeks though.

ancientmooner commented 3 years ago

The code is available now. Sorry for the late release due to refactorization.