Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Concept Videos

Visual System01:26

Visual System

422
Light enters the eye through the cornea, a transparent, dome-shaped surface covering the surface of the eyeball that helps to direct and focus incoming light. This light is then channeled toward the pupil, an adjustable opening whose size is controlled by the iris. The iris, a pigmented muscle, regulates the amount of light entering the eye by contracting or dilating the pupil, thereby ensuring optimal light levels for clear vision.
Once through the pupil, the light passes through the lens, a...
422
Parallel Processing01:20

Parallel Processing

125
The brain processes sensory information rapidly due to parallel processing, which involves sending data across multiple neural pathways at the same time. This method allows the brain to manage various sensory qualities, such as shapes, colors, movements, and locations, all concurrently. For instance, when observing a forest landscape, the brain simultaneously processes the movement of leaves, the shapes of trees, the depth between them, and the various shades of green. This enables a quick and...
125
Vision01:24

Vision

52.2K
Vision is the result of light being detected and transduced into neural signals by the retina of the eye. This information is then further analyzed and interpreted by the brain. First, light enters the front of the eye and is focused by the cornea and lens onto the retina—a thin sheet of neural tissue lining the back of the eye. Because of refraction through the convex lens of the eye, images are projected onto the retina upside-down and reversed.
52.2K

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Nanodrug-Mediated Thermotherapy of Cancer Stem-Like Cells.

Journal of nanoscience and nanotechnology·2016
Same author

Irisin exerts dual effects on browning and adipogenesis of human white adipocytes.

American journal of physiology. Endocrinology and metabolism·2016
Same author

Asymmetric Supercapacitor Based on Porous N-doped Carbon Derived from Pomelo Peel and NiO Arrays.

ACS applied materials & interfaces·2016
Same author

Akt and β-catenin contribute to TMZ resistance and EMT of MGMT negative malignant glioma cell line.

Journal of the neurological sciences·2016
Same author

[Anatomic characteristics of the vessels in the spermatic cord of the varicocele patient: A laparoscopic study].

Zhonghua nan ke xue = National journal of andrology·2016
Same author

Expression of dynein, cytoplasmic 2, heavy chain 1 (DHC2) associated with glioblastoma cell resistance to temozolomide.

Scientific reports·2016

Related Experiment Video

Updated: May 9, 2025

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
03:31

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

428

Dual-stage feature specialization network for robust visual object detection in autonomous vehicles.

Ze Liu1, Junhua Wu2, Yingfeng Cai2

  • 1Automotive Engineering Research Institute, Jiangsu University, Zhenjiang, 212013, Jiangsu, China. liuzee1314@foxmail.com.

Scientific Reports
|May 3, 2025
PubMed
Summary

This study introduces a Dual-Stage Feature Specialization Network (DSFSN) for autonomous vehicle vision. Our method improves object detection accuracy and efficiency, especially for small objects in complex scenes.

Keywords:
Autonomous vehicleComplex environment perceptionComplex problem applicationTwo-stage feature extractionVisual object detection

More Related Videos

Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment
08:25

Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment

Published on: May 7, 2019

8.9K
A Step-by-Step Implementation of DeepBehavior, Deep Learning Toolbox for Automated Behavior Analysis
05:41

A Step-by-Step Implementation of DeepBehavior, Deep Learning Toolbox for Automated Behavior Analysis

Published on: February 6, 2020

9.3K

Related Experiment Videos

Last Updated: May 9, 2025

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
03:31

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

428
Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment
08:25

Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment

Published on: May 7, 2019

8.9K
A Step-by-Step Implementation of DeepBehavior, Deep Learning Toolbox for Automated Behavior Analysis
05:41

A Step-by-Step Implementation of DeepBehavior, Deep Learning Toolbox for Automated Behavior Analysis

Published on: February 6, 2020

9.3K

Area of Science:

  • Computer Vision
  • Machine Learning
  • Autonomous Systems

Background:

  • Feature interference in two-stage object detection hinders performance in complex autonomous driving scenes.
  • Current methods struggle with balancing accuracy and computational efficiency for real-time visual perception.

Purpose of the Study:

  • To develop a novel network, the Dual-Stage Feature Specialization Network (DSFSN), for enhanced visual perception in autonomous vehicles.
  • To address feature interference by decoupling feature extraction for region generation and classification.

Main Methods:

  • Employed MobileNetV3 for efficient candidate region generation.
  • Utilized ResNet-FPN for robust multi-scale feature fusion in classification.
  • Implemented a dual-stage feature specialization approach to mitigate interference.

Main Results:

  • Achieved state-of-the-art performance on PASCAL VOC and MS COCO datasets.
  • Reached 81.6% mAP, outperforming Faster R-CNN by 9.3%, and 29.3% AP on MS COCO.
  • Demonstrated a 14.9% improvement in small object detection accuracy.

Conclusions:

  • The DSFSN offers a robust and efficient framework for autonomous driving visual perception.
  • The proposed method shows significant improvements in accuracy and efficiency, particularly in challenging conditions.
  • Validated through real-world tests in diverse environments like rain and nighttime driving.