相关实验视频

Updated: Sep 14, 2025

Deep Neural Networks for Image-Based Dietary Assessment

Deep Neural Networks for Image-Based Dietary Assessment

Published on: March 13, 2021

FSF-ViT:图像增强和适应性全球-本地特征融合用于Few-Shot食品分类

Jinhong Li¹, Huiying Xu², Xinzhong Zhu²

¹Zhejiang Key Laboratory of Intelligent Education Technology and Application, Zhejiang Normal University, Zhejiang, 321004, China; School of Computer Science and Technology, Zhejiang Normal University, Zhejiang, 321004, China; Information Engineering College, Jinhua University of Vocational Technology, Zhejiang, 321004, China.

|July 19, 2025

概括

我们开发了FSF-ViT,这是一个新的AI模型,用于用有限的数据对食物图像进行分类. 这种方法显著提高了有效的饮食管理和健康跟踪的准确性.

科学领域:

计算机科学计算机科学
人工智能的人工智能
机器学习机器学习

背景情况:

由于数据稀缺,近距离学习对于食品分类至关重要.
传统方法在有限的数据集上扎,增加收集和注释成本.
视觉变压器 (ViT) 模型具有潜力,但需要适应少数拍摄场景.

研究的目的:

引入FSF-ViT,这是一个基于视觉变压器的新型模型,用于Few-Shot Food (FSF) 分类.
为了减少食品图像识别中的数据收集和注释费用.
提高食品分类系统的准确性和效率.

主要方法:

通过整合图像增强和自适应的全球-本地特征融合,开发了FSF-ViT.
使用有限的食物图像数据训练模型.
在一个定制的Food-30数据集和三个基准数据集上评估性能.

主要成果:

在测试组中,FSF-ViT实现了最高分类准确率95.1%.
与标准ViT模型相比,显示了显著的准确性改进 (增加12.8%至15.1%).
通过对分类结果的可视化来验证模型的有效性.

结论:

FSF-ViT提供了一种低成本,高精度的解决方案,用于几次射击的食品分类.
该模型为基于智能设备的饮食记录提供了有效的技术支持.
通过改进的食品识别技术,推进饮食管理和健康监测.

关键词:

中国食品图像数据集深度学习是一种深度学习.饮食管理饮食管理有限的培训样本有限的培训样本低成本的分类技术的技术.

更多相关视频

Design and Evaluation of Smart Glasses for Food Intake and Physical Activity Classification

Design and Evaluation of Smart Glasses for Food Intake and Physical Activity Classification

Published on: February 14, 2018

Integration of Animal Behavioral Assessment and Convolutional Neural Network to Study Wasabi-Alcohol Taste-Smell Interaction

Integration of Animal Behavioral Assessment and Convolutional Neural Network to Study Wasabi-Alcohol Taste-Smell Interaction

Published on: August 16, 2024

相关概念视频

Tagging and Fusion Proteins

Tagging and Fusion Proteins

Proteins are involved in several cellular processes and biochemical reactions. Analyzing a specific protein of interest requires it to be isolated from the other proteins in the cell. This is achieved by overexpressing the specific gene in a suitable host to produce large quantities of the target protein. A tag or label is recombined with the gene to produce a fusion protein containing the target protein and the tag. The tags on these fusion proteins can then be used for easy detection and...

Force Classification

Force Classification

Forces play a crucial role in the study of physics and engineering. They are essential in describing the motion, behavior, and equilibrium of objects in the physical world. Forces can be classified based on their origin, type, and direction of action.
Contact and non-contact forces are two of the most widely used categories of forces. As the name suggests, contact forces require physical contact between two objects to act upon each other. Examples of contact forces include frictional,...