智能检测瓶装液体算法模型的图像训练数据

企业自主采集多环境、多角度、多场景下的瓶装液体的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瓶装液体的算法模型。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瓶装液体。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瓶装液体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，构建形成一个包含瓶装液体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瓶装液体特征，通过循证规则来完成瓶装液体的智能识别，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瓶装液体的智能检测模型。
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。

An enterprise independently collected X-ray security inspection images of bottled liquids under various environments, angles and scenarios, carried out cleaning, annotation and other processing, and used these as samples to train an intelligent detection algorithm model for bottled liquids. This model can be applied to various security inspection scenarios to accurately and quickly detect whether bottled liquids are contained in the inspected objects.
1. Data Source: The original data was collected using self-developed X-ray security inspection equipment, by capturing transmitted images of bottled liquids in various forms under multiple angles and scenarios, to build an original X-ray data atlas library for bottled liquids.
2. Data Processing: Preprocessing operations including geometric transformation, pixel transformation, denoising and matting are performed on the collected raw data. Then, semi-automatic annotation tools are used to generate pseudo-labels for the data, followed by manual correction of the annotations and 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 bottled liquids.
3. Training and Generation Rules of Detection Model: The processed and annotated dataset is imported into a visual detection algorithm model (e.g., Faster R-CNN model) as sample data for deep learning. Through supervised learning, the model learns to identify the features of bottled liquids in the dataset, and completes intelligent recognition of bottled liquids via evidence-based rules, outputting relevant attributes including target item and target position. Furthermore, image attribute information of the detected target objects can be exported, such as image type, image format and acquisition time, etc. The finally generated model is an intelligent detection model capable of accurately identifying bottled liquids.
4. Data Tuning: Hyperparameter tuning is adopted to optimize the model, specifically including learning rate, model structure and size, target loss function and others, so as 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1691条瓶装液体的X光安检图像数据，用于训练智能检测算法模型，支持多环境、多角度、多场景下的瓶装液体检测。