DetectoRS: Detecting Objects with Recursive Feature Pyramid and Switchable Atrous Convolution

Many modern object detectors demonstrate outstanding performances by using the mechanism of looking and thinking twice. In this paper, we explore this mechanism in the backbone design for object detection... (read more)

PDF Abstract

Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK BENCHMARK
Instance Segmentation COCO test-dev DetectoRS (ResNeXt-101-32x4d, multi-scale) mask AP 47.1 # 1
AP50 71.1 # 1
AP75 51.6 # 1
APS 30.3 # 1
APM 49.5 # 2
APL 59.6 # 3
Object Detection COCO test-dev DetectoRS (ResNeXt-101-32x4d, single-scale) box AP 53.3 # 9
AP50 71.6 # 11
AP75 58.5 # 8
APS 33.9 # 13
APM 56.5 # 8
APL 66.9 # 4
Object Detection COCO test-dev DetectoRS (ResNeXt-101-32x4d, multi-scale) box AP 54.7 # 5
AP50 73.5 # 2
AP75 60.1 # 4
APS 37.4 # 3
APM 57.3 # 6
APL 66.4 # 7
Panoptic Segmentation COCO test-dev DetectoRS (ResNeXt-101-32x4d, multi-scale) PQ 49.6 # 1
PQst 37.1 # 2
PQth 57.8 # 1

Methods used in the Paper