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

## TL;DR

This paper introduces NAS-FPN, a neural architecture search-based method to automatically design scalable feature pyramid networks that improve object detection accuracy and efficiency over manually designed models.

## Contribution

The paper presents NAS-FPN, a novel scalable search space and architecture discovered via neural architecture search for feature pyramid networks in object detection.

## Key findings

- NAS-FPN achieves higher detection accuracy than state-of-the-art models.
- NAS-FPN improves mobile detection AP by 2 points.
- NAS-FPN surpasses Mask R-CNN in accuracy with less computation.

## Abstract

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. We adopt Neural Architecture Search and discover a new feature pyramid architecture in a novel scalable search space covering all cross-scale connections. The discovered architecture, named NAS-FPN, consists of a combination of top-down and bottom-up connections to fuse features across scales. NAS-FPN, combined with various backbone models in the RetinaNet framework, achieves better accuracy and latency tradeoff compared to state-of-the-art object detection models. NAS-FPN improves mobile detection accuracy by 2 AP compared to state-of-the-art SSDLite with MobileNetV2 model in [32] and achieves 48.3 AP which surpasses Mask R-CNN [10] detection accuracy with less computation time.