Fod-detr: improved detr based foreign object detection for coal mine conveyor

Yunbo Kang

doi:10.1117/12.3032920

5 July 2024 Fod-detr: improved detr based foreign object detection for coal mine conveyor

Yunbo Kang

Proceedings Volume 13184, Third International Conference on Electronic Information Engineering and Data Processing (EIEDP 2024); 131846A (2024) https://doi.org/10.1117/12.3032920
Event: 3rd International Conference on Electronic Information Engineering and Data Processing (EIEDP 2024), 2024, Kuala Lumpur, Malaysia

Abstract

To address the challenges of morphological diversity and scale changes in foreign object detection of coal mine conveyors, an improved DETR based foreign object detection method FOD-DETR is proposed in this paper. FOD-DETR uses EfficientNet as the backbone feature extraction network. Multi-scale cross axial attention (MSCAA) is incorporated to enhance the feature extraction of multi-scale and variously shaped foreign objects. By introducing multi-scale features into axial attention and establishing dual cross attention between two spatial axial attention, MSCAA leverages multi-scale features for globally identify the ambiguous boundaries. To further address the training convergence issue caused by learnable queries in the Decoder, dynamic anchor boxes are employed as queries for modeling the positional attention maps. Experimental results demonstrate that FOD-DETR achieved a 7.86 improvement in AP50 compared to the vanilla DETR. Among the 13 object detection baselines in comparative experiments, it achieved the highest accuracy, demonstrating the effectiveness of the proposed improvements.

(2024) Published by SPIE. Downloading of the abstract is permitted for personal use only.

Citation Download Citation

Yunbo Kang "Fod-detr: improved detr based foreign object detection for coal mine conveyor", Proc. SPIE 13184, Third International Conference on Electronic Information Engineering and Data Processing (EIEDP 2024), 131846A (5 July 2024); https://doi.org/10.1117/12.3032920

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available

Members: $17.00

Non-members: $21.00 ADD TO CART

PROCEEDINGS
6 PAGES

DOWNLOAD PAPER SAVE TO MY LIBRARY

GET CITATION

RIGHTS & PERMISSIONS

Get copyright permission Get copyright permission on Copyright Marketplace

KEYWORDS

Object detection

Land mines

Transformers

Feature extraction

Education and training

Detection and tracking algorithms

Mining

Show All Keywords

Keywords/Phrases

Search In:

Publication Years