Detection of Remaining Feed in the Feed Troughs of Flat-Fed Meat Ducks Based on the RGB-D Sensor and YOLO V8

The remaining feed in the feed troughs affects the feeding management of flat-raised meat ducks. Ground-contact detection methods all involve modifications to the feeding troughs, but the breeding setting of flat-raised meat ducks does not allow for on-site electrical wiring installation. Additional...

Full description

Saved in:
Bibliographic Details
Main Authors: Xueliang Tan, Junjie Yuan, Shijia Ying, Jizhang Wang
Format: Article
Language:English
Published: MDPI AG 2025-05-01
Series:Animals
Subjects:
Online Access:https://www.mdpi.com/2076-2615/15/10/1440
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850255276651839488
author Xueliang Tan
Junjie Yuan
Shijia Ying
Jizhang Wang
author_facet Xueliang Tan
Junjie Yuan
Shijia Ying
Jizhang Wang
author_sort Xueliang Tan
collection DOAJ
description The remaining feed in the feed troughs affects the feeding management of flat-raised meat ducks. Ground-contact detection methods all involve modifications to the feeding troughs, but the breeding setting of flat-raised meat ducks does not allow for on-site electrical wiring installation. Additionally, the existing non-contact methods do not directly detect the remaining feed quantity in the feeding troughs. To tackle this problem, this study employs a novel approach by first capturing images of the feed troughs using an RGB-D sensor. Subsequently, YOLOv8 is utilized to identify the positions of the feed troughs, and the volume of the remaining feed is determined through point cloud processing. The accuracy of this detection method was evaluated using various types of feed troughs and feed particle sizes. The experimental results reveal both a strong correlation between the calculated and actual feed volumes (with R<sup>2</sup> values exceeding 0.86, indicating a consistent trend) and a low prediction error, as quantified by the root mean square error (RMSE). Analyses of the correction coefficients and corresponding RMSE values indicated a positive correlation between the correction coefficient and the curvature of the feeding trough, while no correlation was observed with the trough diameter or granule particle size, maintaining a low RMSE value. The findings of this research demonstrate the effectiveness of the proposed method for detecting the remaining feed in troughs. This method facilitates precise feed management, minimizes residual feed, and enhances the living conditions of meat ducks.
format Article
id doaj-art-3f0af6ccf7c544949f80e8d253ac3307
institution OA Journals
issn 2076-2615
language English
publishDate 2025-05-01
publisher MDPI AG
record_format Article
series Animals
spelling doaj-art-3f0af6ccf7c544949f80e8d253ac33072025-08-20T01:56:55ZengMDPI AGAnimals2076-26152025-05-011510144010.3390/ani15101440Detection of Remaining Feed in the Feed Troughs of Flat-Fed Meat Ducks Based on the RGB-D Sensor and YOLO V8Xueliang Tan0Junjie Yuan1Shijia Ying2Jizhang Wang3School of Agricultural Engineering, Jiangsu University, Zhenjiang 212013, ChinaSchool of Agricultural Engineering, Jiangsu University, Zhenjiang 212013, ChinaInstitute of Animal Science, Jiangsu Academy of Agricultural Sciences, Nanjing 210014, ChinaSchool of Agricultural Engineering, Jiangsu University, Zhenjiang 212013, ChinaThe remaining feed in the feed troughs affects the feeding management of flat-raised meat ducks. Ground-contact detection methods all involve modifications to the feeding troughs, but the breeding setting of flat-raised meat ducks does not allow for on-site electrical wiring installation. Additionally, the existing non-contact methods do not directly detect the remaining feed quantity in the feeding troughs. To tackle this problem, this study employs a novel approach by first capturing images of the feed troughs using an RGB-D sensor. Subsequently, YOLOv8 is utilized to identify the positions of the feed troughs, and the volume of the remaining feed is determined through point cloud processing. The accuracy of this detection method was evaluated using various types of feed troughs and feed particle sizes. The experimental results reveal both a strong correlation between the calculated and actual feed volumes (with R<sup>2</sup> values exceeding 0.86, indicating a consistent trend) and a low prediction error, as quantified by the root mean square error (RMSE). Analyses of the correction coefficients and corresponding RMSE values indicated a positive correlation between the correction coefficient and the curvature of the feeding trough, while no correlation was observed with the trough diameter or granule particle size, maintaining a low RMSE value. The findings of this research demonstrate the effectiveness of the proposed method for detecting the remaining feed in troughs. This method facilitates precise feed management, minimizes residual feed, and enhances the living conditions of meat ducks.https://www.mdpi.com/2076-2615/15/10/1440localization of feed troughsdeep learningpoint cloud processingvolumetric determination
spellingShingle Xueliang Tan
Junjie Yuan
Shijia Ying
Jizhang Wang
Detection of Remaining Feed in the Feed Troughs of Flat-Fed Meat Ducks Based on the RGB-D Sensor and YOLO V8
Animals
localization of feed troughs
deep learning
point cloud processing
volumetric determination
title Detection of Remaining Feed in the Feed Troughs of Flat-Fed Meat Ducks Based on the RGB-D Sensor and YOLO V8
title_full Detection of Remaining Feed in the Feed Troughs of Flat-Fed Meat Ducks Based on the RGB-D Sensor and YOLO V8
title_fullStr Detection of Remaining Feed in the Feed Troughs of Flat-Fed Meat Ducks Based on the RGB-D Sensor and YOLO V8
title_full_unstemmed Detection of Remaining Feed in the Feed Troughs of Flat-Fed Meat Ducks Based on the RGB-D Sensor and YOLO V8
title_short Detection of Remaining Feed in the Feed Troughs of Flat-Fed Meat Ducks Based on the RGB-D Sensor and YOLO V8
title_sort detection of remaining feed in the feed troughs of flat fed meat ducks based on the rgb d sensor and yolo v8
topic localization of feed troughs
deep learning
point cloud processing
volumetric determination
url https://www.mdpi.com/2076-2615/15/10/1440
work_keys_str_mv AT xueliangtan detectionofremainingfeedinthefeedtroughsofflatfedmeatducksbasedonthergbdsensorandyolov8
AT junjieyuan detectionofremainingfeedinthefeedtroughsofflatfedmeatducksbasedonthergbdsensorandyolov8
AT shijiaying detectionofremainingfeedinthefeedtroughsofflatfedmeatducksbasedonthergbdsensorandyolov8
AT jizhangwang detectionofremainingfeedinthefeedtroughsofflatfedmeatducksbasedonthergbdsensorandyolov8