摘要
一位新闻记者兼机器人与机器学习的新闻编辑每日新闻-机器人的研究发现被用在一份新的报告中。据《中华人民共和国哈尔滨消息》报道，NewsRx记者称，“用餐检测是保证用餐辅助机器人成功率的一项重要技术。然而，由于不同场景下用餐的外观、手势和复杂特征表现出强烈的类间相似性和类内变异性，实时、准确地检测用餐更加困难。”本研究的资金来源包括国家重点研发项目、中央大学基础研究基金。针对上述问题,提出了一种基于可变形卷积法和CloFormer(CF)变压器对YOLOv8s进行优化的新方法,以实现高效、准确的膳食检测,并引入可变形卷积法对YOLOv8s模型架构进行了改进,以获取更细粒度的空间信息。通过共享权重和上下文感知权重来捕获高频局部和低频全局信息,我们将该方法命名为DCF-YOLOv8s。该方法在膳食数据集上进行了实验,分别用基线模型和几个最先进的(SOTA)检测模型对实验数据进行了评价,结果表明,所提出的方法具有较好的性能。该方法在43.6帧/秒和(FPS)帧时平均准确率达到88.5%(mAP)，验证了该方法在餐饮机器人餐饮检测中的有效性和准确性，并通过烧蚀实验验证了引入变形卷积和Clo前处理模块的有效性，验证了采用数据增强方法的重要性。
Abstract
By a News Reporter-Staff News Editor at Robotics & Machine Learning Daily News Daily News – Research findings on Robotics are disc ussed in a new report. According to news originating from Harbin, People’s Repub lic of China, by NewsRx correspondents, research stated, “Meal detection is an i mportant technology to ensure success rate of meal -assisting robotics. However, due to the strong interclass similarity and intraclass variability presented by appearance, gesture, and complex traits of meals in different scenarios, it is more challenging to real-time and accurate detect meals.” Funders for this research include National Key R & D Program of Ch ina, Fundamental Research Funds for the Central Universities. Our news journalists obtained a quote from the research from Harbin Engineering University, “To address the above problems, a novel method based on deformable c onvolution and CloFormer (CF) transformer to optimize the YOLOv8s was proposed t o achieve efficient and accurate detection for meal. The YOLOv8s model architect ure was enhanced by introducing deformable convolution to capture finer -grained spatial information, and the CloFormer module was introduced to capture high -f requency local and low -frequency global information through shared weights and context -aware weights, we notated it as DCF-YOLOv8s. The proposed method was ev aluated on meal datasets, which were evaluated separately with baseline model an d several state-of-the-art (SOTA) detection models, and results show that the pr oposed method achieves better performance. Specifically, the proposed method can achieve 88.5% mean average accuracy (mAP) at 43.6 frames per seco nd (FPS), validating its efficiency and accuracy in meal detection for meal -ass isting robotics. The effectiveness of introducing deformable convolution and Clo Former modules was verified by ablation experiments, and validating the importan ce of adopting data augmentation methods.”
NETL
NSTL