NAS-FPN: Learning Scalable Feature Pyramid Architecture for Object Detection

CVPR 2019 Golnaz GhiasiTsung-Yi LinRuoming PangQuoc V. Le

Current state-of-the-art convolutional architectures for object detection are manually designed. Here we aim to learn a better architecture of feature pyramid network for object detection... (read more)

PDF Abstract

Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT LEADERBOARD
Real-Time Object Detection COCO NAS-FPN AmoebaNet (7 @ 384) + DropBlock MAP 48.3 # 2
FPS 3.6 # 14
inference time (ms) 278.9 # 4
Real-Time Object Detection COCO NAS-FPNLite MobileNetV2 (7 @ 64) MAP 25.7 # 16
inference time (ms) 285 # 5

Methods used in the Paper