智能检测铅蓄电池算法模型的图像训练数据

企业自主采集各类型铅蓄电池的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前蓄电池的的算法模型。该模型可应用在各类安检场景中，针对性精准、快速检测被检物品中是否包含铅蓄电池物品。1、数据来源：原始数据使用自研X光安检设备，多角度、多场景下透射各类型铅蓄电池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，构建形成一个包含铅蓄电池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铅蓄电池特征，通过循证规则来完成铅蓄电池的智能识别，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铅蓄电池的智能检测模型。
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。

Enterprises independently collect X-ray security inspection images of various types of lead-acid batteries, perform processing such as cleaning and annotation, and use these as samples to train an intelligent detection algorithm model for pre-inspection lead-acid batteries. This model can be applied to various security inspection scenarios to accurately and quickly detect whether lead-acid batteries are contained in the inspected items in a targeted manner.
1. Data Source: The original data was collected using self-developed X-ray security inspection equipment, which captures and establishes an original X-ray data legend library of various types of lead-acid batteries through transmission at multiple angles and scenarios.
2. Data Processing: Preprocess the collected original data including geometric transformation, pixel transformation, denoising, matting, etc.; use semi-automatic annotation tools to obtain pseudo-labels, then manually correct the annotations and set an audit mechanism to ensure the accuracy and consistency of annotations, thus constructing a dataset containing X-ray security inspection data of lead-acid batteries.
3. Detection Model Training and Generation Rules: Import the processed and annotated dataset as sample data for deep learning into a visual detection algorithm model (such as the Faster R-CNN model). Let the model learn to recognize the features of lead-acid batteries in the dataset through supervised learning, complete intelligent identification of lead-acid batteries based on evidence-based rules, and output relevant attributes including target item type and target location. Furthermore, the image attribute information of the detected target object can also be exported, such as image type, image format, and acquisition time. The finally generated model is an intelligent detection model that can accurately identify lead-acid batteries.
4. Data Tuning: Optimize the model through hyperparameter tuning, including learning rate, model structure and size, target loss function, etc., to continuously improve the model's detection performance.

搜集汇总

数据集介绍

特点

该数据集包含1375条铅蓄电池的X光安检图像数据，格式为csv，用于训练智能检测铅蓄电池的算法模型，主要应用于安检场景。数据由浙江啄云智能科技有限公司自行产生，并经过清洗、标注等处理，更新频次为半年度更新。