Academic Journal
Προβολή σε EDS
Συνδεδεμένο Πλήρες Κείμενο

PGFD-YOLO: A dual-modal object detection framework with progressive gated fusion and foreground-guided distillation.

Λεπτομέρειες βιβλιογραφικής εγγραφής
Τίτλος: PGFD-YOLO: A dual-modal object detection framework with progressive gated fusion and foreground-guided distillation.
Συγγραφείς: Liu S; School of information, Yunnan Normal University, Kunming, China. Electronic address: 2043205000194@ynnu.edu.cn., Yun L; Department of Education of Yunnan Province, Yunnan Normal University, Kunming, China. Electronic address: yunlijun@ynnu.edu.cn.
Πηγή: Journal of environmental management [J Environ Manage] 2026 Jun 15; Vol. 410, pp. 130053. Date of Electronic Publication: 2026 May 28.
Τύπος έκδοσης: Journal Article
Γλώσσα: English
Στοιχεία περιοδικού: Publisher: Academic Press Country of Publication: England NLM ID: 0401664 Publication Model: Print-Electronic Cited Medium: Internet ISSN: 1095-8630 (Electronic) Linking ISSN: 03014797 NLM ISO Abbreviation: J Environ Manage Subsets: MEDLINE
Imprint Name(s): Original Publication: London ; New York, Academic Press.
Ιατρικοί όροι (MeSH): Waste Management*/methods , Detection Algorithms*
Περίληψη: Effective waste classification is essential for improving resource recovery and reducing landfill burden. However, existing methods predominantly rely on single RGB imaging, which is insufficient for distinguishing overlapping, transparent, or similarly colored waste items. This study proposes a dual-modal waste detection model, PGFD-YOLO, based on YOLOv11, which integrates RGB and depth information without requiring dedicated depth sensors. Depth maps were generated from standard RGB images using monocular depth estimation (Depth Anything V2). Three architectural innovations were introduced: LKC3-F for lightweight dual-stream feature extraction via partial convolution, QAT-PSA for high-level semantic refinement via quad-enhanced attention, and PCGF for stable cross-modal integration via progressive zero-initialized gating. A foreground-guided cross-scale knowledge distillation strategy further improved the accuracy without increasing the inference cost. Experiments on the TACO-10 dataset with 10-fold cross-validation show that PGFD-YOLO achieves mAP50 of 22.3±0.8% and mAP50:95 of 16.5±0.7%, improving over the single-modal baseline by 7.7% and 7.2% absolute, and outperforming naive early and late fusion strategies by 8.1% and 6.4%, respectively. A projected deployment analysis estimates approximately 70%-75% labor cost reduction and $91,700-167,400 annual recyclable recovery value from a medium-scale facility. The model operates at 73.6 FPS, thereby meeting the real-time waste sorting requirements.
(Copyright © 2026 Elsevier Ltd. All rights reserved.)
Competing Interests: Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Contributed Indexing: Keywords: Deep learning; Knowledge distillation; PGFD-YOLO; RGB-D fusion; Waste classification; YOLOv11
Entry Date(s): Date Created: 20260528 Date Completed: 20260613 Latest Revision: 20260621
Update Code: 20260621
DOI: 10.1016/j.jenvman.2026.130053
PMID: 42208255
Βάση Δεδομένων: MEDLINE
FullText Links:
  – Type: other
    Url: https://resolver.ebsco.com:443/public/rma-ftfapi/ejs/direct?AccessToken=485B802143211BC603F4&Show=Object
Text:
  Availability: 0
CustomLinks:
  – Url: https://www.doi.org/10.1016/j.jenvman.2026.130053?
    Name: ScienceDirect (all content) (s7799221)
    Category: fullText
    Text: View record from ScienceDirect
    MouseOverText: View record from ScienceDirect
Header DbId: cmedm
DbLabel: MEDLINE
An: 42208255
AccessLevel: 3
PubType: Academic Journal
PubTypeId: academicJournal
PreciseRelevancyScore: 0
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: PGFD-YOLO: A dual-modal object detection framework with progressive gated fusion and foreground-guided distillation.
– Name: Author
  Label: Authors
  Group: Au
  Data: <searchLink fieldCode="AU" term="%22Liu+S%22">Liu S</searchLink>; School of information, Yunnan Normal University, Kunming, China. Electronic address: 2043205000194@ynnu.edu.cn.<br /><searchLink fieldCode="AU" term="%22Yun+L%22">Yun L</searchLink>; Department of Education of Yunnan Province, Yunnan Normal University, Kunming, China. Electronic address: yunlijun@ynnu.edu.cn.
– Name: TitleSource
  Label: Source
  Group: Src
  Data: <searchLink fieldCode="JN" term="%220401664%22">Journal of environmental management</searchLink> [J Environ Manage] 2026 Jun 15; Vol. 410, pp. 130053. <i>Date of Electronic Publication: </i>2026 May 28.
– Name: TypePub
  Label: Publication Type
  Group: TypPub
  Data: Journal Article
– Name: Language
  Label: Language
  Group: Lang
  Data: English
– Name: TitleSource
  Label: Journal Info
  Group: Src
  Data: <i>Publisher: </i><searchLink fieldCode="PB" term="%22Academic+Press%22">Academic Press </searchLink><i>Country of Publication: </i>England <i>NLM ID: </i>0401664 <i>Publication Model: </i>Print-Electronic <i>Cited Medium: </i>Internet <i>ISSN: </i>1095-8630 (Electronic) <i>Linking ISSN: </i><searchLink fieldCode="IS" term="%2203014797%22">03014797 </searchLink><i>NLM ISO Abbreviation: </i>J Environ Manage <i>Subsets: </i>MEDLINE
– Name: PublisherInfo
  Label: Imprint Name(s)
  Group: PubInfo
  Data: <i>Original Publication</i>: London ; New York, Academic Press.
– Name: SubjectMESH
  Label: MeSH Terms
  Group: Su
  Data: <searchLink fieldCode="MM" term="%22Waste+Management%22">Waste Management*</searchLink>/<searchLink fieldCode="MM" term="%22Waste+Management+methods%22">methods</searchLink> <br /><searchLink fieldCode="MM" term="%22Detection+Algorithms%22">Detection Algorithms*</searchLink>
– Name: Abstract
  Label: Abstract
  Group: Ab
  Data: Effective waste classification is essential for improving resource recovery and reducing landfill burden. However, existing methods predominantly rely on single RGB imaging, which is insufficient for distinguishing overlapping, transparent, or similarly colored waste items. This study proposes a dual-modal waste detection model, PGFD-YOLO, based on YOLOv11, which integrates RGB and depth information without requiring dedicated depth sensors. Depth maps were generated from standard RGB images using monocular depth estimation (Depth Anything V2). Three architectural innovations were introduced: LKC3-F for lightweight dual-stream feature extraction via partial convolution, QAT-PSA for high-level semantic refinement via quad-enhanced attention, and PCGF for stable cross-modal integration via progressive zero-initialized gating. A foreground-guided cross-scale knowledge distillation strategy further improved the accuracy without increasing the inference cost. Experiments on the TACO-10 dataset with 10-fold cross-validation show that PGFD-YOLO achieves mAP<subscript>50</subscript> of 22.3±0.8% and mAP<subscript>50:95</subscript> of 16.5±0.7%, improving over the single-modal baseline by 7.7% and 7.2% absolute, and outperforming naive early and late fusion strategies by 8.1% and 6.4%, respectively. A projected deployment analysis estimates approximately 70%-75% labor cost reduction and $91,700-167,400 annual recyclable recovery value from a medium-scale facility. The model operates at 73.6 FPS, thereby meeting the real-time waste sorting requirements.<br /> (Copyright © 2026 Elsevier Ltd. All rights reserved.)
– Name: Abstract
  Label: Competing Interests
  Group: Ab
  Data: Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
– Name: SubjectMinor
  Label: Contributed Indexing
  Group:
  Data: <i>Keywords: </i>Deep learning; Knowledge distillation; PGFD-YOLO; RGB-D fusion; Waste classification; YOLOv11
– Name: DateEntry
  Label: Entry Date(s)
  Group: Date
  Data: <i>Date Created: </i>20260528 <i>Date Completed: </i>20260613 <i>Latest Revision: </i>20260621
– Name: DateUpdate
  Label: Update Code
  Group: Date
  Data: 20260621
– Name: DOI
  Label: DOI
  Group: ID
  Data: 10.1016/j.jenvman.2026.130053
– Name: AN
  Label: PMID
  Group: ID
  Data: 42208255
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=cmedm&AN=42208255
RecordInfo BibRecord:
  BibEntity:
    Identifiers:
      – Type: doi
        Value: 10.1016/j.jenvman.2026.130053
    Languages:
      – Code: eng
        Text: English
    PhysicalDescription:
      Pagination:
        StartPage: 130053
    Subjects:
      – SubjectFull: Waste Management methods
        Type: general
      – SubjectFull: Detection Algorithms
        Type: general
    Titles:
      – TitleFull: PGFD-YOLO: A dual-modal object detection framework with progressive gated fusion and foreground-guided distillation.
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Liu S
      – PersonEntity:
          Name:
            NameFull: Yun L
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 15
              M: 06
              Text: 2026 Jun 15
              Type: published
              Y: 2026
          Identifiers:
            – Type: issn-electronic
              Value: 1095-8630
          Numbering:
            – Type: volume
              Value: 410
          Titles:
            – TitleFull: Journal of environmental management
              Type: main
ResultId 1