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

Benchmark Model Rank Results
object-detection-on-cocoRelationNet++ (ResNeXt-64x4d-101-DCN)#67box mAP: 52.7