首页 > 最新文献

Smart agricultural technology最新文献

英文 中文
YSD-BPTrack: A multi-object tracking framework for calves in occluded environments
IF 6.3 Q1 AGRICULTURAL ENGINEERING Pub Date : 2025-03-06 DOI: 10.1016/j.atech.2025.100876
Wangli Hao, Chao Ren, Yulong Fan, Meng Han, Fuzhong Li
The accurate tracking of individual calves is essential for health monitoring. However, existing multi tracking frameworks often encounter frequent ID abnormal switching issues during occlusion. To address these challenges, we propose a novel multi-object tracking framework named YSD-BPTrack for calves in occluded environments on cattle farms in this paper. This framework mainly consists of two stages: detection and tracking. Concerning the detection phase, the DCNv4 is integrated into the YOLOv8s model to capture spatial deformation features caused by occlusion, thereby enhancing detection performance under occlusion. Additionally, the Star operation of StarNet is also applied to the model to achieve excellent detection performance with lower computational costs. Concerning the tracking stage, we first propose an innovative rematching algorithm (Rematching module) and a new trajectory removal strategy (Trajectory removal module). The Rematching module performs rematching with detection boxes utilizing extended trajectory prediction boxes in cases of occlusion, resulting in a reduced probability of ID switch errors. Moreover, the Trajectory Removal module dynamically adjusts the removal time for lost matching trajectories, which decreases the likelihood of trajectories being mistakenly removed. Specifically, our proposed novel framework achieves a HOTA (Higher Order Tracking Accuracy) of 91.6%, surpassing other frameworks in both track accuracy and efficiency. Experimental results also validate the superiority of the YSD-BPTrack, with HOTA increasing by 17.6%, MOTA (Multiple Object Tracking Accuracy) by 13.9%, MOTP (Multiple Object Tracking Precision) by 1.8%, IDF1 (Identification F1 Score) by 15.4%, and reducing parameters by 49.1%, IDSw (Identification Switches) by 88.9%, and computational overhead by 39.2% compared to the other frameworks. Overall, the proposed multi-object tracking framework has great potential to revolutionize the tracking of calves.
{"title":"YSD-BPTrack: A multi-object tracking framework for calves in occluded environments","authors":"Wangli Hao,&nbsp;Chao Ren,&nbsp;Yulong Fan,&nbsp;Meng Han,&nbsp;Fuzhong Li","doi":"10.1016/j.atech.2025.100876","DOIUrl":"10.1016/j.atech.2025.100876","url":null,"abstract":"<div><div>The accurate tracking of individual calves is essential for health monitoring. However, existing multi tracking frameworks often encounter frequent ID abnormal switching issues during occlusion. To address these challenges, we propose a novel multi-object tracking framework named YSD-BPTrack for calves in occluded environments on cattle farms in this paper. This framework mainly consists of two stages: detection and tracking. Concerning the detection phase, the DCNv4 is integrated into the YOLOv8s model to capture spatial deformation features caused by occlusion, thereby enhancing detection performance under occlusion. Additionally, the Star operation of StarNet is also applied to the model to achieve excellent detection performance with lower computational costs. Concerning the tracking stage, we first propose an innovative rematching algorithm (Rematching module) and a new trajectory removal strategy (Trajectory removal module). The Rematching module performs rematching with detection boxes utilizing extended trajectory prediction boxes in cases of occlusion, resulting in a reduced probability of ID switch errors. Moreover, the Trajectory Removal module dynamically adjusts the removal time for lost matching trajectories, which decreases the likelihood of trajectories being mistakenly removed. Specifically, our proposed novel framework achieves a HOTA (Higher Order Tracking Accuracy) of 91.6%, surpassing other frameworks in both track accuracy and efficiency. Experimental results also validate the superiority of the YSD-BPTrack, with HOTA increasing by 17.6%, MOTA (Multiple Object Tracking Accuracy) by 13.9%, MOTP (Multiple Object Tracking Precision) by 1.8%, <span><math><mi>ID</mi><msub><mrow><mi>F</mi></mrow><mrow><mn>1</mn></mrow></msub></math></span> (Identification <span><math><msub><mrow><mi>F</mi></mrow><mrow><mn>1</mn></mrow></msub></math></span> Score) by 15.4%, and reducing parameters by 49.1%, IDSw (Identification Switches) by 88.9%, and computational overhead by 39.2% compared to the other frameworks. Overall, the proposed multi-object tracking framework has great potential to revolutionize the tracking of calves.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"11 ","pages":"Article 100876"},"PeriodicalIF":6.3,"publicationDate":"2025-03-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143570588","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
Validation of the FERTI-drip model for the evaluation and simulation of fertigation events in drip irrigation 验证用于滴灌施肥事件评估和模拟的 FERTI-drip 模型
IF 6.3 Q1 AGRICULTURAL ENGINEERING Pub Date : 2025-01-27 DOI: 10.1016/j.atech.2025.100803
Manuel Morcillo, Miguel Ángel Moreno, Rocío Ballesteros, Rocío Arias, José Fernando Ortega
FERTI-DRIP tool was developed to support decision-making in the fertigation process. The model determines the distribution of fertilizers at the subunit level, along with emitter discharge. FERTI-DRIP enables users to optimize the design of the irrigation subunit and the type and discharge of the installed emitter. Additionally, it helps determine the injection, advance, and cleaning times of fertilizer. Despite all these capacities, FERTI-DRIP has not been validated in a real irrigation system. Thus, the objective of this work has been to develop and implement a methodology to evaluate the actual distribution of fertilizers in a commercial productive environment and to calibrate FERTI-DRIP for this purpose.
This methodology has been applied in a commercial farm of 9 ha of almond trees (Prunus dulcis var. avijor), which is equipped with a drip irrigation system, intensively analyzing a fertigation event under real operating conditions.
This study has enabled the validation of the FERTI-DRIP tool, developing a methodology for evaluating fertigation systems that characterizes the distribution of fertilizers and their dynamics during an irrigation event. With the validation of FERTI-DRIP, a real fertigation process has been simulated, achieving a maximum time lag of three minutes and differences in the amount of fertilizer applied by each emitter of <1 %. This enables decision-making regarding fertilizer application times. The model has accurately predicted the dose of fertilizer, the application and cleaning times at the dripper level, thereby improving application efficiencies and mitigating issues arising from salt accumulation in the subunit through proper time management.
{"title":"Validation of the FERTI-drip model for the evaluation and simulation of fertigation events in drip irrigation","authors":"Manuel Morcillo,&nbsp;Miguel Ángel Moreno,&nbsp;Rocío Ballesteros,&nbsp;Rocío Arias,&nbsp;José Fernando Ortega","doi":"10.1016/j.atech.2025.100803","DOIUrl":"10.1016/j.atech.2025.100803","url":null,"abstract":"<div><div>FERTI-DRIP tool was developed to support decision-making in the fertigation process. The model determines the distribution of fertilizers at the subunit level, along with emitter discharge. FERTI-DRIP enables users to optimize the design of the irrigation subunit and the type and discharge of the installed emitter. Additionally, it helps determine the injection, advance, and cleaning times of fertilizer. Despite all these capacities, FERTI-DRIP has not been validated in a real irrigation system. Thus, the objective of this work has been to develop and implement a methodology to evaluate the actual distribution of fertilizers in a commercial productive environment and to calibrate FERTI-DRIP for this purpose.</div><div>This methodology has been applied in a commercial farm of 9 ha of almond trees (<em>Prunus dulcis</em> var. avijor), which is equipped with a drip irrigation system, intensively analyzing a fertigation event under real operating conditions.</div><div>This study has enabled the validation of the FERTI-DRIP tool, developing a methodology for evaluating fertigation systems that characterizes the distribution of fertilizers and their dynamics during an irrigation event. With the validation of FERTI-DRIP, a real fertigation process has been simulated, achieving a maximum time lag of three minutes and differences in the amount of fertilizer applied by each emitter of &lt;1 %. This enables decision-making regarding fertilizer application times. The model has accurately predicted the dose of fertilizer, the application and cleaning times at the dripper level, thereby improving application efficiencies and mitigating issues arising from salt accumulation in the subunit through proper time management.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"10 ","pages":"Article 100803"},"PeriodicalIF":6.3,"publicationDate":"2025-01-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143182146","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
Spectral bands vs. vegetation indices: An AutoML approach for processing tomato yield predictions based on Sentinel-2 imagery
IF 6.3 Q1 AGRICULTURAL ENGINEERING Pub Date : 2025-01-25 DOI: 10.1016/j.atech.2025.100805
Nicoleta Darra , Borja Espejo-Garcia , Vassilis Psiroukis , Emmanouil Psomiadis , Spyros Fountas
Predicting tomato yield is crucial for agricultural production planning, management, market supply, and risk management. While yield prediction in precision agriculture presents a complex challenge, advancements in remote sensor technologies have greatly improved the accuracy and feasibility of these predictions. In this paper, three modeling approaches were implemented to estimate processing tomato yield based on satellite data, covering 152 fields across two growing periods. Sentinel-2 spectral bands and Vegetation Indices (VIs) captured at 5-day intervals during the growth periods, used as input parameters in an AutoML pipeline. The first approach aimed to identify the optimal timeframe and the most effective spectral bands and VIs, such as NDVI and RVI. Modelling results indicated that Red/Red Edge/NIR bands performed well in predicting yield, with the period between 75 to 90 days post-transplanting identified as the optimal timeframe for yieldpredictions (R² of 0.56). The second approach incorporated inter-date VIs, utilizing bands from different dates, leading to a significant improvement in performance with an R² of 0.61 and root mean square error (RMSE) of 12ton/ha. The third approach involved band combinations to enhance performance, where specific bands, including the Bands 4, 6, and 12, collectively achieved the highest R² of 0.65. Using feature extraction algorithms such as PCA, UMAP, and autoencoder partially contributed to improved performance while using the same VIs on consecutive different dates. By utilizing a higher number of bands and dates without the constraint of a VI formula demonstrated the potential for enhanced model accuracy.
{"title":"Spectral bands vs. vegetation indices: An AutoML approach for processing tomato yield predictions based on Sentinel-2 imagery","authors":"Nicoleta Darra ,&nbsp;Borja Espejo-Garcia ,&nbsp;Vassilis Psiroukis ,&nbsp;Emmanouil Psomiadis ,&nbsp;Spyros Fountas","doi":"10.1016/j.atech.2025.100805","DOIUrl":"10.1016/j.atech.2025.100805","url":null,"abstract":"<div><div>Predicting tomato yield is crucial for agricultural production planning, management, market supply, and risk management. While yield prediction in precision agriculture presents a complex challenge, advancements in remote sensor technologies have greatly improved the accuracy and feasibility of these predictions. In this paper, three modeling approaches were implemented to estimate processing tomato yield based on satellite data, covering 152 fields across two growing periods. Sentinel-2 spectral bands and Vegetation Indices (VIs) captured at 5-day intervals during the growth periods, used as input parameters in an AutoML pipeline. The first approach aimed to identify the optimal timeframe and the most effective spectral bands and VIs, such as NDVI and RVI. Modelling results indicated that Red/Red Edge/NIR bands performed well in predicting yield, with the period between 75 to 90 days post-transplanting identified as the optimal timeframe for yieldpredictions (R² of 0.56). The second approach incorporated inter-date VIs, utilizing bands from different dates, leading to a significant improvement in performance with an R² of 0.61 and root mean square error (RMSE) of 12ton/ha. The third approach involved band combinations to enhance performance, where specific bands, including the Bands 4, 6, and 12, collectively achieved the highest R² of 0.65. Using feature extraction algorithms such as PCA, UMAP, and autoencoder partially contributed to improved performance while using the same VIs on consecutive different dates. By utilizing a higher number of bands and dates without the constraint of a VI formula demonstrated the potential for enhanced model accuracy.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"10 ","pages":"Article 100805"},"PeriodicalIF":6.3,"publicationDate":"2025-01-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143182188","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
Factors influencing learning attitude of farmers regarding adoption of farming technologies in farms of Kentucky, USA
IF 6.3 Q1 AGRICULTURAL ENGINEERING Pub Date : 2025-01-23 DOI: 10.1016/j.atech.2025.100801
Dipesh Oli, Buddhi Gyawali, Shikha Acharya, Samuel Oshikoya
Understanding farmers’ learning attitudes towards agricultural technologies and the factors affecting their adoption behavior is crucial in today's era of rapid technological advancement. This study focuses on assessing various socioeconomic factors influencing farmers’ learning behavior regarding technologies in farming operations and understanding farmers’ views on the future of precision agriculture in crop and livestock production. The study also examined the preferred methods farmers use for learning. Using R-studio, the binary logistic regression model was employed to analyze the survey data. The results suggest that the education level and social media use significantly affected farmers’ learning attitudes. In contrast factors such as gender, age, income level, related expertise, and farming experience had no significant impact. The study also concluded that seminars, workshops, and training are preferred learning methods. Thus, it is recommended that federal and state agencies and universities' extension systems should focus on combining these preferred learning methods with various social media platforms to disseminate the necessary information, knowledge, and skills to farmers, supporting better adoption of agricultural technologies.
{"title":"Factors influencing learning attitude of farmers regarding adoption of farming technologies in farms of Kentucky, USA","authors":"Dipesh Oli,&nbsp;Buddhi Gyawali,&nbsp;Shikha Acharya,&nbsp;Samuel Oshikoya","doi":"10.1016/j.atech.2025.100801","DOIUrl":"10.1016/j.atech.2025.100801","url":null,"abstract":"<div><div>Understanding farmers’ learning attitudes towards agricultural technologies and the factors affecting their adoption behavior is crucial in today's era of rapid technological advancement. This study focuses on assessing various socioeconomic factors influencing farmers’ learning behavior regarding technologies in farming operations and understanding farmers’ views on the future of precision agriculture in crop and livestock production. The study also examined the preferred methods farmers use for learning. Using R-studio, the binary logistic regression model was employed to analyze the survey data. The results suggest that the education level and social media use significantly affected farmers’ learning attitudes. In contrast factors such as gender, age, income level, related expertise, and farming experience had no significant impact. The study also concluded that seminars, workshops, and training are preferred learning methods. Thus, it is recommended that federal and state agencies and universities' extension systems should focus on combining these preferred learning methods with various social media platforms to disseminate the necessary information, knowledge, and skills to farmers, supporting better adoption of agricultural technologies.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"10 ","pages":"Article 100801"},"PeriodicalIF":6.3,"publicationDate":"2025-01-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143182186","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
Precision agriculture for iceberg lettuce: From spatial sensing to per plant decision making and control
IF 6.3 Q1 AGRICULTURAL ENGINEERING Pub Date : 2025-01-22 DOI: 10.1016/j.atech.2025.100797
William Rohde, Fulvio Forni
Precision agriculture enables growers to sense growth in the field and take actions at a high resolution. In this context, agriculture can be considered a feedback system, and feedback control algorithms are a promising approach for decision-making at a per-plant level. We present a control approach to assign per-plant nitrogen prescriptions for iceberg lettuce, with the objective of reducing variability at harvest and, as a result, increasing yield. The per-plant nitrogen inputs were generated using a proportional consensus-protocol style controller and non-destructive measurements of growth. This resulted in smaller lettuce plants receiving an increased dose of nitrogen and oversized plants receiving a decreased dose. In this paper, we present the results of three in-field trials applying the controller to real plants with manual applications. Our results demonstrated a reduction in the plant mass variance of outdoor crop in two of our three farm trials (32.6% and 19.7% lower variance in comparison to cohorts with a typical approach). This is a promising result, which could be improved in future work by developing a more accurate proxy measurement for crop growth and accounting for soil nitrogen.
{"title":"Precision agriculture for iceberg lettuce: From spatial sensing to per plant decision making and control","authors":"William Rohde,&nbsp;Fulvio Forni","doi":"10.1016/j.atech.2025.100797","DOIUrl":"10.1016/j.atech.2025.100797","url":null,"abstract":"<div><div>Precision agriculture enables growers to sense growth in the field and take actions at a high resolution. In this context, agriculture can be considered a feedback system, and feedback control algorithms are a promising approach for decision-making at a per-plant level. We present a control approach to assign per-plant nitrogen prescriptions for iceberg lettuce, with the objective of reducing variability at harvest and, as a result, increasing yield. The per-plant nitrogen inputs were generated using a proportional consensus-protocol style controller and non-destructive measurements of growth. This resulted in smaller lettuce plants receiving an increased dose of nitrogen and oversized plants receiving a decreased dose. In this paper, we present the results of three in-field trials applying the controller to real plants with manual applications. Our results demonstrated a reduction in the plant mass variance of outdoor crop in two of our three farm trials (32.6% and 19.7% lower variance in comparison to cohorts with a typical approach). This is a promising result, which could be improved in future work by developing a more accurate proxy measurement for crop growth and accounting for soil nitrogen.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"10 ","pages":"Article 100797"},"PeriodicalIF":6.3,"publicationDate":"2025-01-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143182133","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
Soil and crop response to varying planter's downforce in corn and cotton fields
IF 6.3 Q1 AGRICULTURAL ENGINEERING Pub Date : 2025-01-22 DOI: 10.1016/j.atech.2025.100798
Luan Oliveira , Brenda Ortiz , Gregory Pate , Thomas Way , Rouverson Silva
Effective planter downforce is crucial for optimizing seed placement and improving crop emergence and growth. This study explores the impact of different downforce settings on soil compaction, seeding depth, and crop performance for corn and cotton. Two field experiments were implemented using a John Deere 6145R tractor and a 6-row John Deere planter equipped with a DeltaForce® hydraulic downforce system in Shorter, Alabama, during the 2020 growing season. Multiple planter's downforce levels (0, 550, 1100, and 1800 N) were tested in both static and dynamic modes across two fields with distinct soil types (clay loam and sandy loam). Results indicated that the dynamic downforce mode provided more accurate and consistent load distribution compared to the static mode, which often exceeded target loads. Increased downforce led to deeper seeding depths, particularly with the dynamic mode, and higher loads in the static mode resulted in greater variability. For corn in clay loam soil, higher static downforce improved seed-to-soil contact, enhancing emergence and plant height. Conversely, cotton in sandy loam soil showed no significant differences in emergence or plant height, likely due to soil moisture conditions and target depth. The study concludes that dynamic downforce systems offer superior load control and uniformity, enhancing corn emergence and growth in heavier soils, while further research is recommended to optimize settings for various crops and soil conditions.
{"title":"Soil and crop response to varying planter's downforce in corn and cotton fields","authors":"Luan Oliveira ,&nbsp;Brenda Ortiz ,&nbsp;Gregory Pate ,&nbsp;Thomas Way ,&nbsp;Rouverson Silva","doi":"10.1016/j.atech.2025.100798","DOIUrl":"10.1016/j.atech.2025.100798","url":null,"abstract":"<div><div>Effective planter downforce is crucial for optimizing seed placement and improving crop emergence and growth. This study explores the impact of different downforce settings on soil compaction, seeding depth, and crop performance for corn and cotton. Two field experiments were implemented using a John Deere 6145R tractor and a 6-row John Deere planter equipped with a DeltaForce® hydraulic downforce system in Shorter, Alabama, during the 2020 growing season. Multiple planter's downforce levels (0, 550, 1100, and 1800 N) were tested in both static and dynamic modes across two fields with distinct soil types (clay loam and sandy loam). Results indicated that the dynamic downforce mode provided more accurate and consistent load distribution compared to the static mode, which often exceeded target loads. Increased downforce led to deeper seeding depths, particularly with the dynamic mode, and higher loads in the static mode resulted in greater variability. For corn in clay loam soil, higher static downforce improved seed-to-soil contact, enhancing emergence and plant height. Conversely, cotton in sandy loam soil showed no significant differences in emergence or plant height, likely due to soil moisture conditions and target depth. The study concludes that dynamic downforce systems offer superior load control and uniformity, enhancing corn emergence and growth in heavier soils, while further research is recommended to optimize settings for various crops and soil conditions.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"10 ","pages":"Article 100798"},"PeriodicalIF":6.3,"publicationDate":"2025-01-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143182187","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
PCCNet: A point supervised dense Chickens flock counting network
IF 6.3 Q1 AGRICULTURAL ENGINEERING Pub Date : 2025-01-21 DOI: 10.1016/j.atech.2025.100795
Yubin Guo , Zhipeng Wu , Zhiwei Su , Jiangsan Zhao , Ximing Li
In broiler breeding, precise counting is crucial for improving production efficiency and ensuring animal welfare. Nevertheless, counting chickens precisely is a challenging task especially when young chicks always huddle for warmth. Although deep learning has been widely taken in different counting related tasks, more accurate localization and counting of chickens in high stocking density scenes still has not been well investigated. We propose a point supervised dense chickens flock counting network (PCCNet), which directly utilizes points as learning targets. The network adopts information feature fusion to assist the identification of broilers high stocking density scenes. In addition, considering the distance of neighboring points as matching cost in point matching algorithms is advantageous for generating more reasonable matching results, facilitating model convergence. To validate the effectiveness of the proposed network, a Chicken Counting Dataset (CCD) is built, consisting of two subsets separated by different ages: CCD_A and CCD_B. The accuracies of PCCNet on the two subsets of CCD are 97.85% and 97.06%, with corresponding Mean Absolute Errors (MAE) of 1.966 and 5.173, and Root Mean Square Errors (RMSE) values of 3.474 and 7.034, respectively. Our model achieves better broiler counting performance than other state-of-the-art (SOTA) methods.
在肉鸡饲养中,精确计数对于提高生产效率和确保动物福利至关重要。然而,对鸡进行精确计数是一项极具挑战性的任务,尤其是当幼雏总是挤在一起取暖时。虽然深度学习已被广泛应用于不同的计数相关任务中,但在高密度放养场景中更精确的鸡只定位和计数仍未得到很好的研究。我们提出了一种直接利用点作为学习目标的点监督密集鸡群计数网络(PCCNet)。该网络采用信息特征融合技术来辅助识别肉鸡高存栏密度场景。此外,在点匹配算法中考虑相邻点的距离作为匹配代价,有利于产生更合理的匹配结果,促进模型收敛。为了验证所提网络的有效性,我们建立了一个鸡计数数据集(CCD),该数据集由两个按不同年龄划分的子集组成:CCD_A 和 CCD_B。PCCNet 在 CCD 两个子集中的准确率分别为 97.85% 和 97.06%,相应的平均绝对误差 (MAE) 分别为 1.966 和 5.173,均方根误差 (RMSE) 分别为 3.474 和 7.034。与其他最先进的(SOTA)方法相比,我们的模型实现了更好的肉鸡计数性能。
{"title":"PCCNet: A point supervised dense Chickens flock counting network","authors":"Yubin Guo ,&nbsp;Zhipeng Wu ,&nbsp;Zhiwei Su ,&nbsp;Jiangsan Zhao ,&nbsp;Ximing Li","doi":"10.1016/j.atech.2025.100795","DOIUrl":"10.1016/j.atech.2025.100795","url":null,"abstract":"<div><div>In broiler breeding, precise counting is crucial for improving production efficiency and ensuring animal welfare. Nevertheless, counting chickens precisely is a challenging task especially when young chicks always huddle for warmth. Although deep learning has been widely taken in different counting related tasks, more accurate localization and counting of chickens in high stocking density scenes still has not been well investigated. We propose a point supervised dense chickens flock counting network (PCCNet), which directly utilizes points as learning targets. The network adopts information feature fusion to assist the identification of broilers high stocking density scenes. In addition, considering the distance of neighboring points as matching cost in point matching algorithms is advantageous for generating more reasonable matching results, facilitating model convergence. To validate the effectiveness of the proposed network, a Chicken Counting Dataset (CCD) is built, consisting of two subsets separated by different ages: CCD_A and CCD_B. The accuracies of PCCNet on the two subsets of CCD are 97.85% and 97.06%, with corresponding Mean Absolute Errors (MAE) of 1.966 and 5.173, and Root Mean Square Errors (RMSE) values of 3.474 and 7.034, respectively. Our model achieves better broiler counting performance than other state-of-the-art (SOTA) methods.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"10 ","pages":"Article 100795"},"PeriodicalIF":6.3,"publicationDate":"2025-01-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143182136","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
Enhancing the Performance of YOLOv9t Through a Knowledge Distillation Approach for Real-Time Detection of Bloomed Damask Roses in the Field
IF 6.3 Q1 AGRICULTURAL ENGINEERING Pub Date : 2025-01-20 DOI: 10.1016/j.atech.2025.100794
Farhad Fatehi, Hossein Bagherpour, Jafar Amiri Parian
Harvesting Damask roses by hand is especially challenging because of the thorns on their stems, which not only complicate the process but also pose a risk of injury to workers. This problem highlights the need for automation solutions to facilitate the harvesting process. To carry out agricultural operation, particularly for picking fully bloomed Damask roses, using harvesting robots offers significant potential to reduce labor costs while simultaneously improving crop quality. Recent developments in deep learning algorithms, especially in convolutional models, have shown significant promise for object detection, highlighting strong possibilities for improving the efficiency of this process. The substantial computational demands and processing times associated with many deep learning models present a significant obstacle to their implementation in real-time applications. To address this challenge, knowledge distillation (KD) has emerged as a valuable model compression technique. This approach enables complex "teacher" models to pass essential insights to more streamlined "student" models, making them more suitable for immediate, real-world applications. In this study, we trained YOLOv9t model as a student model and YOLOv9c model as a teacher model. To facilitate this learning, two different techniques including online distillation (OD) and offline distillation (OFD) were explored. The results demonstrated that applying both online and offline KD increased the mAP0.5 of YOLOv9t by 0.3% and 0.2%, respectively, and boosted the detection speed by 5.1 and 1.8 frames per second (FPS), respectively. The results showed that the YOLOv9t model, trained as a student with both OD and OFD methods, performed better than the YOLOv9t model. This distilled version of YOLOv9t shows strong potential as an effective model for real-time detection of fully bloomed Damask roses.
人工采摘大马士革玫瑰尤其具有挑战性,因为它们的茎上有刺,这不仅使采摘过程变得复杂,而且还会给工人带来受伤的风险。这一问题凸显了采用自动化解决方案来促进采摘过程的必要性。在进行农业作业,特别是采摘完全盛开的大马士革玫瑰时,使用采摘机器人可以大大降低劳动力成本,同时提高作物质量。深度学习算法的最新发展,特别是卷积模型的发展,已在物体检测方面显示出巨大的前景,为提高这一过程的效率提供了巨大的可能性。与许多深度学习模型相关的大量计算需求和处理时间对其在实时应用中的实施构成了重大障碍。为了应对这一挑战,知识蒸馏(KD)已成为一种有价值的模型压缩技术。这种方法能让复杂的 "教师 "模型将重要的见解传递给更精简的 "学生 "模型,使它们更适合即时的真实世界应用。在本研究中,我们将 YOLOv9t 模型作为学生模型进行训练,将 YOLOv9c 模型作为教师模型进行训练。为了促进这种学习,我们探索了两种不同的技术,包括在线蒸馏(OD)和离线蒸馏(OFD)。结果表明,应用在线和离线 KD,YOLOv9t 的 mAP0.5 分别提高了 0.3% 和 0.2%,检测速度分别提高了 5.1 帧/秒和 1.8 帧/秒 (FPS)。结果表明,以学生身份同时接受 OD 和 OFD 方法训练的 YOLOv9t 模型比 YOLOv9t 模型表现更好。这个经过提炼的 YOLOv9t 版本显示出作为实时检测盛开的大马士革玫瑰的有效模型的强大潜力。
{"title":"Enhancing the Performance of YOLOv9t Through a Knowledge Distillation Approach for Real-Time Detection of Bloomed Damask Roses in the Field","authors":"Farhad Fatehi,&nbsp;Hossein Bagherpour,&nbsp;Jafar Amiri Parian","doi":"10.1016/j.atech.2025.100794","DOIUrl":"10.1016/j.atech.2025.100794","url":null,"abstract":"<div><div>Harvesting Damask roses by hand is especially challenging because of the thorns on their stems, which not only complicate the process but also pose a risk of injury to workers. This problem highlights the need for automation solutions to facilitate the harvesting process. To carry out agricultural operation, particularly for picking fully bloomed Damask roses, using harvesting robots offers significant potential to reduce labor costs while simultaneously improving crop quality. Recent developments in deep learning algorithms, especially in convolutional models, have shown significant promise for object detection, highlighting strong possibilities for improving the efficiency of this process. The substantial computational demands and processing times associated with many deep learning models present a significant obstacle to their implementation in real-time applications. To address this challenge, knowledge distillation (KD) has emerged as a valuable model compression technique. This approach enables complex \"teacher\" models to pass essential insights to more streamlined \"student\" models, making them more suitable for immediate, real-world applications. In this study, we trained YOLOv9t model as a student model and YOLOv9c model as a teacher model. To facilitate this learning, two different techniques including online distillation (OD) and offline distillation (OFD) were explored. The results demonstrated that applying both online and offline KD increased the mAP0.5 of YOLOv9t by 0.3% and 0.2%, respectively, and boosted the detection speed by 5.1 and 1.8 frames per second (FPS), respectively. The results showed that the YOLOv9t model, trained as a student with both OD and OFD methods, performed better than the YOLOv9t model. This distilled version of YOLOv9t shows strong potential as an effective model for real-time detection of fully bloomed Damask roses.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"10 ","pages":"Article 100794"},"PeriodicalIF":6.3,"publicationDate":"2025-01-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143182132","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
Automated precision weighing: Leveraging 2D video feature analysis and machine learning for live body weight estimation of broiler chickens
IF 6.3 Q1 AGRICULTURAL ENGINEERING Pub Date : 2025-01-19 DOI: 10.1016/j.atech.2025.100793
Mairead Campbell , Paul Miller , Katerine Díaz-Chito , Sean Irvine , Mary Baxter , Jesús Martínez Del Rincón , Xin Hong , Niall McLaughlin , Thianantha Arumugam , Niamh O'Connell
The measurement of bird live weight during the production cycle is an important management practice in commercial broiler farming. However, the accuracy and practicalities of current weighing methods are limited. This paper proposes a non-invasive system that uses low-cost, overhead cameras combined with computer vision and AI techniques to automatically weigh broiler chickens. The main objectives were to: (i) evaluate 2D video feature descriptors, together with regression modelling, to predict the live weight of broilers; (ii) establish the impact of posture (i.e. sitting/standing) and bird age on the accuracy of weight estimation; (iii) assess the feasibility of the camera-weighing system to monitor weight at different bird ages. In the first experiment, a video feature analysis was performed to evaluate the accuracy of 2D feature descriptors (ellipse axes, ellipse area, bounding box width, bounding box height) to predict the weight of broilers. Individual birds were manually weighed to establish a reference weight. The relationship between the feature sets and the reference weight was evaluated using six multivariate regression models. The approach was tested on two groups of broilers aged 23 (n = 21 broilers) and 35 (n = 23 broilers) days old, weighing between 570 and 2980 g. In experiment 2, the best performing feature set and linear regression modelling from experiment 1 were applied to a larger number of birds across a greater age range (5 to 35 days old, n = 222 broilers). To be more representative of the intended application of this technology, footage was recorded from the feeding area of a commercial broiler house and an automated chicken detector and tracking method was applied. The model was retrained using reference weights from experiment 2 (ranging from 100 to 3085 g) to refine model performance. In experiment 1, the posture feature did not improve weight estimation whilst age improved the performance of all models. The accuracy of body weight estimation was greatest when bird age and the minor ellipse axis (x,y endpoints of the maximum points that are perpendicular to the longest line that can be drawn through an object) were used as model features. In experiment 2, the model showed the poorest performance in 5-day old birds with a mean relative error of 12.1 ± 7.9 %. Overall, the model could estimate the weight of a broiler chicken with a mean relative error of 7.0 ± 5.8 %. The results indicate that the analysis of 2D image features using video analytics and regression modelling is a promising method of obtaining rapid, cost-effective and accurate estimates of broiler live weight.
{"title":"Automated precision weighing: Leveraging 2D video feature analysis and machine learning for live body weight estimation of broiler chickens","authors":"Mairead Campbell ,&nbsp;Paul Miller ,&nbsp;Katerine Díaz-Chito ,&nbsp;Sean Irvine ,&nbsp;Mary Baxter ,&nbsp;Jesús Martínez Del Rincón ,&nbsp;Xin Hong ,&nbsp;Niall McLaughlin ,&nbsp;Thianantha Arumugam ,&nbsp;Niamh O'Connell","doi":"10.1016/j.atech.2025.100793","DOIUrl":"10.1016/j.atech.2025.100793","url":null,"abstract":"<div><div>The measurement of bird live weight during the production cycle is an important management practice in commercial broiler farming. However, the accuracy and practicalities of current weighing methods are limited. This paper proposes a non-invasive system that uses low-cost, overhead cameras combined with computer vision and AI techniques to automatically weigh broiler chickens. The main objectives were to: (i) evaluate 2D video feature descriptors, together with regression modelling, to predict the live weight of broilers; (ii) establish the impact of posture (i.e. sitting/standing) and bird age on the accuracy of weight estimation; (iii) assess the feasibility of the camera-weighing system to monitor weight at different bird ages. In the first experiment, a video feature analysis was performed to evaluate the accuracy of 2D feature descriptors (ellipse axes, ellipse area, bounding box width, bounding box height) to predict the weight of broilers. Individual birds were manually weighed to establish a reference weight. The relationship between the feature sets and the reference weight was evaluated using six multivariate regression models. The approach was tested on two groups of broilers aged 23 (<em>n</em> = 21 broilers) and 35 (<em>n</em> = 23 broilers) days old, weighing between 570 and 2980 g. In experiment 2, the best performing feature set and linear regression modelling from experiment 1 were applied to a larger number of birds across a greater age range (5 to 35 days old, <em>n</em> = 222 broilers). To be more representative of the intended application of this technology, footage was recorded from the feeding area of a commercial broiler house and an automated chicken detector and tracking method was applied. The model was retrained using reference weights from experiment 2 (ranging from 100 to 3085 g) to refine model performance. In experiment 1, the posture feature did not improve weight estimation whilst age improved the performance of all models. The accuracy of body weight estimation was greatest when bird age and the <em>minor ellipse axis</em> (x,y endpoints of the maximum points that are perpendicular to the longest line that can be drawn through an object) were used as model features. In experiment 2, the model showed the poorest performance in 5-day old birds with a mean relative error of 12.1 ± 7.9 %. Overall, the model could estimate the weight of a broiler chicken with a mean relative error of 7.0 ± 5.8 %. The results indicate that the analysis of 2D image features using video analytics and regression modelling is a promising method of obtaining rapid, cost-effective and accurate estimates of broiler live weight.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"10 ","pages":"Article 100793"},"PeriodicalIF":6.3,"publicationDate":"2025-01-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143182183","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
Research on kiwifruit harvesting robot worldwide: A solution for sustainable development of kiwifruit industry
IF 6.3 Q1 AGRICULTURAL ENGINEERING Pub Date : 2025-01-19 DOI: 10.1016/j.atech.2025.100792
Zhiwei Tian , Xiangyu Guo , Wei Ma , Xinyu Xue
Kiwifruit harvesting is labor-intensive, and social issues like an aging population and a declining agricultural workforce have significantly increased costs, presenting unprecedented challenges to the industry. Automatic harvesting systems utilizing multi-sensor fusion, AI, and automation technologies show great potential for replacing manual labor in kiwi harvesting. This paper reviews over 140 research articles related to kiwi fruit harvesting robots, summarizing existing progress in two key areas: target fruit recognition and positioning systems, and fruit picking and collection systems. We compare the pros and cons of various methods, including traditional image recognition and deep learning, active and passive localization techniques, diverse end-effector design structure and driving mechanisms, robotic arm path planning, and harvesting systems. The results show that challenges remain in the commercialization of kiwi harvesting robots. The absence of a unified evaluation standard for robot performance makes the latest research achievements hard to be inherited, leading to slow advancements. Current algorithms are often not lightweight enough for low-cost embedded systems. Additionally, the reliance on manual labeling of dense targets and the accumulation of system error compromise the robustness of target recognition and spatial positioning in open environments. The existing studies tend to focus on local improvements rather than the entire harvesting system. So addressing these issues should be a priority for future research. This paper can provide a reference for researchers and assist industry professionals in understanding the trends in harvesting robot development.
{"title":"Research on kiwifruit harvesting robot worldwide: A solution for sustainable development of kiwifruit industry","authors":"Zhiwei Tian ,&nbsp;Xiangyu Guo ,&nbsp;Wei Ma ,&nbsp;Xinyu Xue","doi":"10.1016/j.atech.2025.100792","DOIUrl":"10.1016/j.atech.2025.100792","url":null,"abstract":"<div><div>Kiwifruit harvesting is labor-intensive, and social issues like an aging population and a declining agricultural workforce have significantly increased costs, presenting unprecedented challenges to the industry. Automatic harvesting systems utilizing multi-sensor fusion, AI, and automation technologies show great potential for replacing manual labor in kiwi harvesting. This paper reviews over 140 research articles related to kiwi fruit harvesting robots, summarizing existing progress in two key areas: target fruit recognition and positioning systems, and fruit picking and collection systems. We compare the pros and cons of various methods, including traditional image recognition and deep learning, active and passive localization techniques, diverse end-effector design structure and driving mechanisms, robotic arm path planning, and harvesting systems. The results show that challenges remain in the commercialization of kiwi harvesting robots. The absence of a unified evaluation standard for robot performance makes the latest research achievements hard to be inherited, leading to slow advancements. Current algorithms are often not lightweight enough for low-cost embedded systems. Additionally, the reliance on manual labeling of dense targets and the accumulation of system error compromise the robustness of target recognition and spatial positioning in open environments. The existing studies tend to focus on local improvements rather than the entire harvesting system. So addressing these issues should be a priority for future research. This paper can provide a reference for researchers and assist industry professionals in understanding the trends in harvesting robot development.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"10 ","pages":"Article 100792"},"PeriodicalIF":6.3,"publicationDate":"2025-01-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143182147","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
期刊
Smart agricultural technology
全部 Acc. Chem. Res. ACS Applied Bio Materials ACS Appl. Electron. Mater. ACS Appl. Energy Mater. ACS Appl. Mater. Interfaces ACS Appl. Nano Mater. ACS Appl. Polym. Mater. ACS BIOMATER-SCI ENG ACS Catal. ACS Cent. Sci. ACS Chem. Biol. ACS Chemical Health & Safety ACS Chem. Neurosci. ACS Comb. Sci. ACS Earth Space Chem. ACS Energy Lett. ACS Infect. Dis. ACS Macro Lett. ACS Mater. Lett. ACS Med. Chem. Lett. ACS Nano ACS Omega ACS Photonics ACS Sens. ACS Sustainable Chem. Eng. ACS Synth. Biol. Anal. Chem. BIOCHEMISTRY-US Bioconjugate Chem. BIOMACROMOLECULES Chem. Res. Toxicol. Chem. Rev. Chem. Mater. CRYST GROWTH DES ENERG FUEL Environ. Sci. Technol. Environ. Sci. Technol. Lett. Eur. J. Inorg. Chem. IND ENG CHEM RES Inorg. Chem. J. Agric. Food. Chem. J. Chem. Eng. Data J. Chem. Educ. J. Chem. Inf. Model. J. Chem. Theory Comput. J. Med. Chem. J. Nat. Prod. J PROTEOME RES J. Am. Chem. Soc. LANGMUIR MACROMOLECULES Mol. Pharmaceutics Nano Lett. Org. Lett. ORG PROCESS RES DEV ORGANOMETALLICS J. Org. Chem. J. Phys. Chem. J. Phys. Chem. A J. Phys. Chem. B J. Phys. Chem. C J. Phys. Chem. Lett. Analyst Anal. Methods Biomater. Sci. Catal. Sci. Technol. Chem. Commun. Chem. Soc. Rev. CHEM EDUC RES PRACT CRYSTENGCOMM Dalton Trans. Energy Environ. Sci. ENVIRON SCI-NANO ENVIRON SCI-PROC IMP ENVIRON SCI-WAT RES Faraday Discuss. Food Funct. Green Chem. Inorg. Chem. Front. Integr. Biol. J. Anal. At. Spectrom. J. Mater. Chem. A J. Mater. Chem. B J. Mater. Chem. C Lab Chip Mater. Chem. Front. Mater. Horiz. MEDCHEMCOMM Metallomics Mol. Biosyst. Mol. Syst. Des. Eng. Nanoscale Nanoscale Horiz. Nat. Prod. Rep. New J. Chem. Org. Biomol. Chem. Org. Chem. Front. PHOTOCH PHOTOBIO SCI PCCP Polym. Chem.
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
0
微信
客服QQ
Book学术公众号 扫码关注我们
反馈
×
意见反馈
请填写您的意见或建议
请填写您的手机或邮箱
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
现在去查看 取消
×
提示
确定
Book学术官方微信
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术
文献互助 智能选刊 最新文献 互助须知 联系我们:info@booksci.cn
Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。
Copyright © 2023 Book学术 All rights reserved.
ghs 京公网安备 11010802042870号 京ICP备2023020795号-1