RetinaMask: Learning to predict masks improves state-of-the-art single-shot detection for free

Recently two-stage detectors have surged ahead of single-shot detectors in the accuracy-vs-speed trade-off. Nevertheless single-shot detectors are immensely popular in embedded vision applications... (read more)

PDF Abstract

Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT BENCHMARK
Object Detection COCO minival RetinaMask (ResNet-101-FPN) box AP 41.1 # 38
AP50 60.2 # 32
AP75 44.1 # 30
Object Detection COCO test-dev RetinaMask (ResNet-50-FPN) box AP 39.4 # 73
AP50 58.6 # 75
AP75 42.3 # 79
APS 21.9 # 71
APM 42.0 # 72
APL 51.0 # 70
Object Detection COCO test-dev RetinaMask (ResNeXt-101-FPN-GN) box AP 42.6 # 56
AP50 62.5 # 54
AP75 46.0 # 61
APS 24.8 # 55
APM 45.6 # 56
APL 53.8 # 56

Methods used in the Paper