An enhanced feature pyramid object detection network for autonomous driving

Yutian Wu, Shuming Tang, Shuwei Zhang, Harutoshi Ogai

Research output: Contribution to journalArticle

Abstract

Feature Pyramid Network (FPN) builds a high-level semantic feature pyramid and detects objects of different scales in corresponding pyramid levels. Usually, features within the same pyramid levels have the same weight for subsequent object detection, which ignores the feature requirements of different scale objects. As we know, for most detection networks, it is hard to detect small objects and occluded objects because there is little information to exploit. To solve the above problems, we propose an Enhanced Feature Pyramid Object Detection Network (EFPN), which innovatively constructs an enhanced feature extraction subnet and adaptive parallel detection subnet. Enhanced feature extraction subnet introduces Feature Weight Module (FWM) to enhance pyramid features by weighting the fusion feature map. Adaptive parallel detection subnet introduces Adaptive Context Expansion (ACE) and Parallel Detection Branch (PDB). ACE aims to generate the features of adaptively enlarged object context region and original region. PDB predicts classification and regression results separately with the two features. Experiments showed that EFPN outperforms FPN in detection accuracy on Pascal VOC and KITTI datasets. Furthermore, the performance of EFPN meets the real-time requirements of autonomous driving systems.

Original languageEnglish
Article number4363
JournalApplied Sciences (Switzerland)
Volume9
Issue number20
DOIs
Publication statusPublished - 2019 Oct 1

Fingerprint

pyramids
Feature extraction
Volatile organic compounds
Fusion reactions
pattern recognition
Semantics
Object detection
requirements
expansion
semantics
volatile organic compounds
Experiments
regression analysis
modules
fusion

Keywords

  • Augmented reality
  • Autonomous driving systems
  • Context embedding
  • Feature pyramid network
  • Feature recalibration
  • Object detection

ASJC Scopus subject areas

  • Materials Science(all)
  • Instrumentation
  • Engineering(all)
  • Process Chemistry and Technology
  • Computer Science Applications
  • Fluid Flow and Transfer Processes

Cite this

An enhanced feature pyramid object detection network for autonomous driving. / Wu, Yutian; Tang, Shuming; Zhang, Shuwei; Ogai, Harutoshi.

In: Applied Sciences (Switzerland), Vol. 9, No. 20, 4363, 01.10.2019.

Research output: Contribution to journalArticle

@article{b9de9af706a340b98b74c456651d74fb,
title = "An enhanced feature pyramid object detection network for autonomous driving",
abstract = "Feature Pyramid Network (FPN) builds a high-level semantic feature pyramid and detects objects of different scales in corresponding pyramid levels. Usually, features within the same pyramid levels have the same weight for subsequent object detection, which ignores the feature requirements of different scale objects. As we know, for most detection networks, it is hard to detect small objects and occluded objects because there is little information to exploit. To solve the above problems, we propose an Enhanced Feature Pyramid Object Detection Network (EFPN), which innovatively constructs an enhanced feature extraction subnet and adaptive parallel detection subnet. Enhanced feature extraction subnet introduces Feature Weight Module (FWM) to enhance pyramid features by weighting the fusion feature map. Adaptive parallel detection subnet introduces Adaptive Context Expansion (ACE) and Parallel Detection Branch (PDB). ACE aims to generate the features of adaptively enlarged object context region and original region. PDB predicts classification and regression results separately with the two features. Experiments showed that EFPN outperforms FPN in detection accuracy on Pascal VOC and KITTI datasets. Furthermore, the performance of EFPN meets the real-time requirements of autonomous driving systems.",
keywords = "Augmented reality, Autonomous driving systems, Context embedding, Feature pyramid network, Feature recalibration, Object detection",
author = "Yutian Wu and Shuming Tang and Shuwei Zhang and Harutoshi Ogai",
year = "2019",
month = "10",
day = "1",
doi = "10.3390/app9204363",
language = "English",
volume = "9",
journal = "Applied Sciences (Switzerland)",
issn = "2076-3417",
publisher = "Multidisciplinary Digital Publishing Institute (MDPI)",
number = "20",

}

TY - JOUR

T1 - An enhanced feature pyramid object detection network for autonomous driving

AU - Wu, Yutian

AU - Tang, Shuming

AU - Zhang, Shuwei

AU - Ogai, Harutoshi

PY - 2019/10/1

Y1 - 2019/10/1

N2 - Feature Pyramid Network (FPN) builds a high-level semantic feature pyramid and detects objects of different scales in corresponding pyramid levels. Usually, features within the same pyramid levels have the same weight for subsequent object detection, which ignores the feature requirements of different scale objects. As we know, for most detection networks, it is hard to detect small objects and occluded objects because there is little information to exploit. To solve the above problems, we propose an Enhanced Feature Pyramid Object Detection Network (EFPN), which innovatively constructs an enhanced feature extraction subnet and adaptive parallel detection subnet. Enhanced feature extraction subnet introduces Feature Weight Module (FWM) to enhance pyramid features by weighting the fusion feature map. Adaptive parallel detection subnet introduces Adaptive Context Expansion (ACE) and Parallel Detection Branch (PDB). ACE aims to generate the features of adaptively enlarged object context region and original region. PDB predicts classification and regression results separately with the two features. Experiments showed that EFPN outperforms FPN in detection accuracy on Pascal VOC and KITTI datasets. Furthermore, the performance of EFPN meets the real-time requirements of autonomous driving systems.

AB - Feature Pyramid Network (FPN) builds a high-level semantic feature pyramid and detects objects of different scales in corresponding pyramid levels. Usually, features within the same pyramid levels have the same weight for subsequent object detection, which ignores the feature requirements of different scale objects. As we know, for most detection networks, it is hard to detect small objects and occluded objects because there is little information to exploit. To solve the above problems, we propose an Enhanced Feature Pyramid Object Detection Network (EFPN), which innovatively constructs an enhanced feature extraction subnet and adaptive parallel detection subnet. Enhanced feature extraction subnet introduces Feature Weight Module (FWM) to enhance pyramid features by weighting the fusion feature map. Adaptive parallel detection subnet introduces Adaptive Context Expansion (ACE) and Parallel Detection Branch (PDB). ACE aims to generate the features of adaptively enlarged object context region and original region. PDB predicts classification and regression results separately with the two features. Experiments showed that EFPN outperforms FPN in detection accuracy on Pascal VOC and KITTI datasets. Furthermore, the performance of EFPN meets the real-time requirements of autonomous driving systems.

KW - Augmented reality

KW - Autonomous driving systems

KW - Context embedding

KW - Feature pyramid network

KW - Feature recalibration

KW - Object detection

UR - http://www.scopus.com/inward/record.url?scp=85074205450&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85074205450&partnerID=8YFLogxK

U2 - 10.3390/app9204363

DO - 10.3390/app9204363

M3 - Article

AN - SCOPUS:85074205450

VL - 9

JO - Applied Sciences (Switzerland)

JF - Applied Sciences (Switzerland)

SN - 2076-3417

IS - 20

M1 - 4363

ER -