M2Det: A Single-Shot Object Detector based on Multi-Level Feature Pyramid Network

Benchmark Model Rank Results
object-detection-on-cocoM2Det (VGG-16, multi-scale)#146box mAP: 44.2AP50: 64.6AP75: 49.3APS: 29.2APM: 47.9APL: 55.1
object-detection-on-cocoM2Det (ResNet-101, multi-scale)#148box mAP: 43.9AP50: 64.4AP75: 48APS: 29.6APM: 49.6APL: 54.3
object-detection-on-cocoM2Det (VGG-16, single-scale)#180box mAP: 41.0AP50: 59.7AP75: 45APS: 22.1APM: 46.5APL: 53.8
object-detection-on-cocoM2Det (ResNet-101, single-scale)#206box mAP: 38.8AP50: 59.4AP75: 41.7APS: 20.5APM: 43.9APL: 53.4
object-detection-on-coco-minivalM2Det (ResNet-1o1, 320x320)#208box AP: 34.1AP50: 53.7APS: 15.9APM: 39.5APL: 49.3
object-detection-on-coco-minivalM2Det (VGG-16, 320x320)#210box AP: 33.2AP50: 52.2APS: 15APM: 38.2APL: 49.1