Search research articles

Related Concept Videos

You might also read

Related Articles

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

Sort by

Same author

Defect-Mask2Former: An Improved Semantic Segmentation Model for Precise Small-Sized Defect Detection on Large-Sized Timbers.

Sensors (Basel, Switzerland)·2026

Same author

Research on Intelligent Wood Species Identification Method Based on Multimodal Texture-Dominated Features and Deep Learning Fusion.

Plants (Basel, Switzerland)·2026

Same author

Intelligent Identification of Tea Plant Seedlings Under High-Temperature Conditions via YOLOv11-MEIP Model Based on Chlorophyll Fluorescence Imaging.

Plants (Basel, Switzerland)·2025

Same author

Characteristic Changes and Potential Markers of Flavour in Raw Pu-Erh Tea with Different Ageing Cycles Analysed by HPLC, HS-SPME-GC-MS, and OAV.

Foods (Basel, Switzerland)·2025

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Video

Updated: Apr 15, 2026

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

FSD-YOLO: A Fusion Framework for Region Segmentation and Deformable Object Detection in Container Yards.

Linghao Dai¹, Zhihong Liang¹, Qi Feng¹

¹College of Big Data and Intelligent Engineering, Southwest Forestry University, Kunming 650224, China.

Sensors (Basel, Switzerland)

|April 14, 2026

Summary

This summary is machine-generated.

This study introduces FSD-YOLO, a unified framework for enhanced safety monitoring in container yards. It improves detection of small and deformable objects, crucial for real-time industrial safety management.

Keywords:

SegFormer YOLOv8 deformable convolution dynamic loss weighting object detection safety monitoring

More Related Videos

From Voxels to Knowledge: A Practical Guide to the Segmentation of Complex Electron Microscopy 3D-Data

From Voxels to Knowledge: A Practical Guide to the Segmentation of Complex Electron Microscopy 3D-Data

Published on: August 13, 2014

Automated Analysis of C. elegans Fluorescence Images using SegElegans

Automated Analysis of C. elegans Fluorescence Images using SegElegans

Published on: October 10, 2025

Related Experiment Videos

Last Updated: Apr 15, 2026

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

From Voxels to Knowledge: A Practical Guide to the Segmentation of Complex Electron Microscopy 3D-Data

From Voxels to Knowledge: A Practical Guide to the Segmentation of Complex Electron Microscopy 3D-Data

Published on: August 13, 2014

Automated Analysis of C. elegans Fluorescence Images using SegElegans

Automated Analysis of C. elegans Fluorescence Images using SegElegans

Published on: October 10, 2025

Area of Science:

Computer Vision
Industrial Safety
Logistics Technology

Background:

Container hoisting operations in rail-road intermodal logistics parks present complex visual challenges.
Existing single-task models struggle with small targets, deformable objects, and occlusions, limiting real-time safety applications.
High-precision perception is essential for preventing accidents in dynamic industrial environments.

Purpose of the Study:

To develop a unified visual analysis framework integrating semantic segmentation and object detection.
To enhance the recognition of small and deformable targets in complex container yard environments.
To enable real-time perception and safety warning for critical objects and hazardous regions.

Main Methods:

Proposed FSD-YOLO architecture combining SegFormer-based semantic segmentation and an improved YOLOv8n object detection network.
Incorporated C2f modules, C2fDCN modules with deformable convolution, and CARAFE upsampling for enhanced feature extraction and fusion.
Implemented a dynamic loss-weighting strategy for small objects and a decision-level fusion of segmentation and detection outputs.

Main Results:

The FSD-YOLO model achieved superior performance on a custom container yard dataset, with mAP50-95 of 0.6433 and mAP50 of 0.9565.
Significantly outperformed the baseline YOLOv8n model (mAP50-95: 0.5394, mAP50: 0.8435).
Demonstrated effectiveness in improving the detection of small and deformable objects in challenging conditions.

Conclusions:

The unified framework effectively addresses limitations of conventional models in complex industrial environments.
The proposed FSD-YOLO architecture enhances real-time perception and safety warning capabilities for container logistics.
The fusion strategy provides robust safety judgments based on semantic rules and object detection.