Segmenting plantar pressure images intelligently can provide valuable insight for people with high blood pressure, making bespoke footwear requirements possible and resulting in more comfortable shoe designs. It is, however, difficult to extract design elements from a segmented image dataset. To address this challenge, we propose an ML-GNN model that segments plantar pressure images using metal-earning. The first part of the paper presents a method for extracting image features that reduce the complexity of the ML-GNN algorithm. To create the network structure, we propose optimization meta-based learning. Using a meta-learning-based graphic neural network, we enhance our mask-based CNN prediction model with VGG16 and CNN layers. We pre-processed the plantar pressure dataset using pressure-sensing data acquisition and compared the results. By defining standard image segmentation indices, we demonstrate the high effectiveness of our research. We have developed an ML-GNN model that improves the segmentation accuracy of plantar pressure images and can also be applied to other sensor image datasets. Through our shoe-last customization approach, we enable the shoe industry to manufacture shoes more efficiently, particularly for people with specific healthcare needs who require bespoke shoe designs. Our findings demonstrate the potential of intelligent image segmentation to advance the field of footwear design and improve the lives of people with specific health requirements.

中文翻译：

---

采用基于元学习的图形卷积神经网络进行足底压力成像的设计元素提取

智能分割足底压力图像可以为高血压患者提供有价值的见解，使定制鞋类需求成为可能，并产生更舒适的鞋类设计。然而，从分割的图像数据集中提取设计元素是很困难的。为了应对这一挑战，我们提出了一种 ML-GNN 模型，该模型使用金属收益来分割足底压力图像。论文第一部分提出了一种提取图像特征的方法，以降低 ML-GNN 算法的复杂度。为了创建网络结构，我们提出了基于元的优化学习。使用基于元学习的图形神经网络，我们通过 VGG16 和 CNN 层增强了基于掩模的 CNN 预测模型。我们使用压力传感数据采集对足底压力数据集进行预处理并比较结果。通过定义标准图像分割指数，我们证明了我们研究的高效性。我们开发了一种 ML-GNN 模型，可以提高足底压力图像的分割精度，也可以应用于其他传感器图像数据集。通过我们的鞋楦定制方法，我们使制鞋行业能够更高效地制造鞋子，特别是对于需要定制鞋子设计、有特定医疗保健需求的人们。我们的研究结果证明了智能图像分割在推进鞋类设计领域并改善有特定健康要求的人们的生活方面的潜力。