Yuxian Pang , Chuchu Chen , Yuedong Yang, Delin Mo
{"title":"Porkolor:猪肉颜色分类的深度学习框架。","authors":"Yuxian Pang , Chuchu Chen , Yuedong Yang, Delin Mo","doi":"10.1016/j.meatsci.2024.109731","DOIUrl":null,"url":null,"abstract":"<div><div>Pork color is crucial for assessing its safety and freshness, and traditional methods of observing through human eyes are inefficient and subjective. In recent years, several methods have been proposed based on computer vision and deep learning have been proposed, which can provide objective and stable evaluations. However, these methods suffer from a lack of standardized data collection methods and large-scale datasets for training, leading to poor model performance and limited generalization capabilities. Additionally, the model accuracy was limited by an absence of effective image preprocessing of background noises.To address these issues, we have designed a standardized pork image collection device and collected 1707 high-quality pork images. Base on the data, we proposed a novel deep learning model to predict the color. The framework consists of two modules: image preprocessing module and pork color classification module. The image preprocessing module uses the Segment Anything Model (SAM) to extract the pork portion and remove background noise, thereby enhancing the model's accuracy and stability. The pork color classification module uses the ResNet-101 model trained with a patch-based training strategy as the backbone. As a result, the model achieved a classification accuracy of 91.50 % on our high quality dataset and 89.00 % on the external validation dataset. The Porkolor online application is freely available at <span><span>https://bio-web1.nscc-gz.cn/app/Porkolor</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":389,"journal":{"name":"Meat Science","volume":"221 ","pages":"Article 109731"},"PeriodicalIF":7.1000,"publicationDate":"2024-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Porkolor: A deep learning framework for pork color classification\",\"authors\":\"Yuxian Pang , Chuchu Chen , Yuedong Yang, Delin Mo\",\"doi\":\"10.1016/j.meatsci.2024.109731\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div><div>Pork color is crucial for assessing its safety and freshness, and traditional methods of observing through human eyes are inefficient and subjective. In recent years, several methods have been proposed based on computer vision and deep learning have been proposed, which can provide objective and stable evaluations. However, these methods suffer from a lack of standardized data collection methods and large-scale datasets for training, leading to poor model performance and limited generalization capabilities. Additionally, the model accuracy was limited by an absence of effective image preprocessing of background noises.To address these issues, we have designed a standardized pork image collection device and collected 1707 high-quality pork images. Base on the data, we proposed a novel deep learning model to predict the color. The framework consists of two modules: image preprocessing module and pork color classification module. The image preprocessing module uses the Segment Anything Model (SAM) to extract the pork portion and remove background noise, thereby enhancing the model's accuracy and stability. The pork color classification module uses the ResNet-101 model trained with a patch-based training strategy as the backbone. As a result, the model achieved a classification accuracy of 91.50 % on our high quality dataset and 89.00 % on the external validation dataset. The Porkolor online application is freely available at <span><span>https://bio-web1.nscc-gz.cn/app/Porkolor</span><svg><path></path></svg></span>.</div></div>\",\"PeriodicalId\":389,\"journal\":{\"name\":\"Meat Science\",\"volume\":\"221 \",\"pages\":\"Article 109731\"},\"PeriodicalIF\":7.1000,\"publicationDate\":\"2024-12-12\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Meat Science\",\"FirstCategoryId\":\"97\",\"ListUrlMain\":\"https://www.sciencedirect.com/science/article/pii/S0309174024003085\",\"RegionNum\":1,\"RegionCategory\":\"农林科学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"Agricultural and Biological Sciences\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Meat Science","FirstCategoryId":"97","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0309174024003085","RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"Agricultural and Biological Sciences","Score":null,"Total":0}
Porkolor: A deep learning framework for pork color classification
Pork color is crucial for assessing its safety and freshness, and traditional methods of observing through human eyes are inefficient and subjective. In recent years, several methods have been proposed based on computer vision and deep learning have been proposed, which can provide objective and stable evaluations. However, these methods suffer from a lack of standardized data collection methods and large-scale datasets for training, leading to poor model performance and limited generalization capabilities. Additionally, the model accuracy was limited by an absence of effective image preprocessing of background noises.To address these issues, we have designed a standardized pork image collection device and collected 1707 high-quality pork images. Base on the data, we proposed a novel deep learning model to predict the color. The framework consists of two modules: image preprocessing module and pork color classification module. The image preprocessing module uses the Segment Anything Model (SAM) to extract the pork portion and remove background noise, thereby enhancing the model's accuracy and stability. The pork color classification module uses the ResNet-101 model trained with a patch-based training strategy as the backbone. As a result, the model achieved a classification accuracy of 91.50 % on our high quality dataset and 89.00 % on the external validation dataset. The Porkolor online application is freely available at https://bio-web1.nscc-gz.cn/app/Porkolor.
期刊介绍:
The aim of Meat Science is to serve as a suitable platform for the dissemination of interdisciplinary and international knowledge on all factors influencing the properties of meat. While the journal primarily focuses on the flesh of mammals, contributions related to poultry will be considered if they enhance the overall understanding of the relationship between muscle nature and meat quality post mortem. Additionally, papers on large birds (e.g., emus, ostriches) as well as wild-captured mammals and crocodiles will be welcomed.