Scholarlink

The global horticulture industry is plagued by endemic issues of efficient sorting and freshness detection of vegetables and fruits. Each human inspection process is labor-intensive, inexact and not precise due to diverse light, surface texture and background noises. Such inefficiencies lead to increased post-harvest loss, lower product quality and lower supply chain performance. This review methodically reviews eighteen peer-reviewed research studies chosen based on a formal search across databases like IEEE Xplore, Scopus, and Web of Science using specified keywords and relevance criteria which brings out essential advances in overcoming issues like environmental heterogeneity, dataset restrictions, and the computational feasibility of real-time deployment. By way of response, recent innovations have turned to deep learning techniques, in the form of convolutional neural networks (CNNs) such as VGG-16, ResNet, and YOLO iterations, to highly automate the grading and classification process with great precision. Such models, coupled with robotic handling systems namely grippers that are fitted with visual and tactile sensors encompass real-time accurate sorting and handling of fruits and vegetables. Literature reviewed in this paper points out notable progress toward overcoming challenges of environmental heterogeneity, data limitations in datasets and computational demands of real-time processing. Approaches like sparsely parameterized model structures, edge optimization for computing and multi-sensor fusion are being explored to improve scalability and adaptability. Also mentioned as critical to broader application is the use of standard datasets and the development of models that support multiple products. This survey indicates potential directions for future investigation in adaptive learning, multimodal integration and sustainable design driven by emerging technologies like hyperspectral imaging and Industry 4.0-based methodologies in the fore at the creation of autonomous, efficient and environmentally sustainable agriculture logistics systems.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Comprehensive Review of Deep Learning and Robotic Automation for Fruit and Vegetable Sorting and Freshness Detection

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Shenila Ansari,
  • Pratik Handore,
  • M. N. Chandan,
  • Vaishnavi Gaikwad,
  • Shweta Desai,
  • Anand Shivashimpi

摘要

The global horticulture industry is plagued by endemic issues of efficient sorting and freshness detection of vegetables and fruits. Each human inspection process is labor-intensive, inexact and not precise due to diverse light, surface texture and background noises. Such inefficiencies lead to increased post-harvest loss, lower product quality and lower supply chain performance. This review methodically reviews eighteen peer-reviewed research studies chosen based on a formal search across databases like IEEE Xplore, Scopus, and Web of Science using specified keywords and relevance criteria which brings out essential advances in overcoming issues like environmental heterogeneity, dataset restrictions, and the computational feasibility of real-time deployment. By way of response, recent innovations have turned to deep learning techniques, in the form of convolutional neural networks (CNNs) such as VGG-16, ResNet, and YOLO iterations, to highly automate the grading and classification process with great precision. Such models, coupled with robotic handling systems namely grippers that are fitted with visual and tactile sensors encompass real-time accurate sorting and handling of fruits and vegetables. Literature reviewed in this paper points out notable progress toward overcoming challenges of environmental heterogeneity, data limitations in datasets and computational demands of real-time processing. Approaches like sparsely parameterized model structures, edge optimization for computing and multi-sensor fusion are being explored to improve scalability and adaptability. Also mentioned as critical to broader application is the use of standard datasets and the development of models that support multiple products. This survey indicates potential directions for future investigation in adaptive learning, multimodal integration and sustainable design driven by emerging technologies like hyperspectral imaging and Industry 4.0-based methodologies in the fore at the creation of autonomous, efficient and environmentally sustainable agriculture logistics systems.