|
| 基于PP-YOLOE+的虾米异物检测系统设计与实现 |
| Design and Implementation of a Foreign Body DetectionSSystemSfor Shrimp Based on PP-YOLOE+ |
| 投稿时间：2024-09-15 修订日期：2025-01-20 |
| DOI： |
| 中文关键词: 虾米，异物检测，异物剔除，PP-YOLOE+ |
| 英文关键词: shrimp, foreign object detection, remove foreign objects, PP-YOLOE+ |
| 基金项目:安徽省高校自然科学研究重点项目(2022AH050252) |
|
| 摘要点击次数: 23 |
| 全文下载次数: 0 |
| 中文摘要: |
| 虾米中混入异物，降低了虾米的质量和经济价值。由于异物在颜色与形态上与虾米十分相似，致使人工目测难以检测剔除异物。为了剔除虾米中的异物，本文设计实现了基于PP-YOLOE+的虾米异物检测系统。系统通过现场图片采集模块采集带有鼠妇等异物的图片，将其标注为训练数据集。在训练数据集上训练几种流行的目标检测模型，对比分析这些目标检测模型的性能，选择PP-YOLOE+作为目标检测模型，并通过计算设计异物剔除模块，根据检测到的异物信息剔除异物。基于PP-YOLOE+的虾米异物检测系统异物识别精度达到88.6%，帧率5.56 FPS，符合并超过系统需求，相对于人工目测检测，降低了人工成本，提高了检测准确度，提升了虾米相关食品质量。 |
| 英文摘要: |
| The inclusion of foreign objects in shrimp reduces its quality and economic value. Due to the similarity in color and shape between foreign objects and shrimp, it is difficult for manual visual inspection to detect and remove foreign objects. To remove foreign objects from shrimp, this article designs and implements a shrimp foreign object detection system based on the PP-YOLOE+. The system collects images of foreign objects such as woodlice through the on-site image acquisition module and annotates them as a training dataset. Several popular object detection models are trained on the training dataset, the performance of these object detection models is compared and analyzed, the PP-YOLOE+ is chosen as the object detection model, and a foreign objects removal module through calculation to remove foreign objects based on the detected foreign object information. The foreign object recognition accuracy of the shrimp foreign object detection system based on the PP-YOLOE+ reaches 88.6%, with a frame rate of 5.56 FPS, which meets and exceeds the system requirements. Compared with manual visual inspection, it reduces labor costs, improves detection accuracy, and enhances shrimp-related food quality. |
|
View Fulltext
查看/发表评论 下载PDF阅读器 |
| 关闭 |
