智能检测电子烟算法模型的图像训练数据

企业自主采集多环境、多角度、多形态的电子烟的X光安检图像，进行清洗、标注等处理，并以此为样本训练生成智能检测电子烟的算法模型。该模型可应用在各类安检场景中，精准、快速检测被检物中是否包电子烟物品。1、数据来源：原始数据使用自研X光安检设备，多角度、多场景下透射各形态的电子烟采集并建立其原始的X光数据图例库。
2、数据处理：对收集到的原始数据进行进行包括几何变换、像素变换、去噪、抠图等预处理；并对数据利用半自动标注工具标注得到伪标签，然后使用人工修正标注，并设置审核机制，保证标注的准确性和一致性，构建形成一个包含电子烟X光安检数据的数据集。
3、检测模型训练生成规则：将处理及标注好的数据集作为深度学习的样本数据导入视觉检测算法模型（如：FasterRCNN模型），通过监督学习的方式让模型学习识别数据集中电子烟特征，通过循证规则来完成电子烟的智能识别，并输出相关属性，包括目标品项、目标位置。进一步的还可将被检目标对象的图像属性信息导出，如图像类型、图像格式以及采集时间等，最终生成的模型为可精准识别电子烟的智能检测模型。
4、数据调优：选择超参数调优的方式对模型优化，具体的包括学习率、模型结构和尺寸、目标损失函数等，持续提升模型检测性能。

Independently collected by the enterprise, X-ray security inspection images of e-cigarettes in various environments, angles and forms are cleaned, annotated and otherwise processed, and used as samples to train and develop an intelligent detection algorithm model for e-cigarettes. This model can be applied in various security inspection scenarios to accurately and quickly detect whether e-cigarettes are contained in the inspected items.
1. Data Source: The original data was collected using self-developed X-ray security inspection equipment, by capturing transmitted X-ray images of e-cigarettes in various forms across multiple angles and scenarios, to establish an original X-ray data image library for e-cigarettes.
2. Data Processing: The collected raw data undergoes preprocessing including geometric transformation, pixel transformation, denoising, matting and other operations; then pseudo-labels are obtained by annotating the data with semi-automatic annotation tools, followed by manual correction of the annotations and the establishment of an audit mechanism to ensure the accuracy and consistency of the annotations, thus constructing a dataset containing X-ray security inspection data of e-cigarettes.
3. Training and Generation Rules of the Detection Model: The processed and annotated dataset is imported into visual detection algorithm models (e.g., Faster R-CNN model) as sample data for deep learning. Through supervised learning, the model learns to recognize the features of e-cigarettes in the dataset, completes intelligent identification of e-cigarettes via evidence-based rules, and outputs relevant attributes including target item and target location. Furthermore, image attribute information of the inspected target objects can be exported, such as image type, image format and acquisition time, etc. The final generated model is an intelligent detection model capable of accurately identifying e-cigarettes.
4. Data Tuning: The model is optimized through hyperparameter tuning, including learning rate, model structure and size, target loss function and other aspects, to continuously improve the detection performance of the model.

搜集汇总

数据集介绍

特点

该数据集是由浙江啄云智能科技有限公司自行产生的电子烟X光安检图像训练数据，包含917条csv格式数据，用于训练智能检测电子烟的算法模型，适用于安检场景中的电子烟检测。