智能检测手铐算法模型的训练数据
收藏资源简介：
本数据知识产权包含多角度、多场景下的手铐的X光安检图像数据，通过对图像的标注、抠图、增强、融合等处理，可作为优质样本训练生成智能检测手铐的算法模型，实现对藏匿在其他物品中或伪装成其他物品等情境下的手铐精准识别。生成的智能检测模型可应用在各类安检场景中1、数据来源：应用X射线光源多角度、多场景下透射手铐，采集并建立其原始的X光数据图例库。 2、数据深度处理：对采集到的原始X光图像预标注坐标位置和品项类别，并对手铐的图像进行抠图处理。将抠出的图像与多场景图像分别进行几何变换、像素变换等增广处理。 3、检测模型生成算法规则：将处理后的手铐图像和场景图像通过密度统计（像素值代表实物密度值）依据区域匹配原则进行融合，融合区域掩模作为数据标签与融合后的图像作为深度学习样本数据。还可通过调整抠图区域在场景图像区域的位置，获得不同的平均密度差值，训练生成可精准定位、精准识别手铐的智能检测模型。区域匹配原则按照Mask*(α*ρ抠图图像+β*ρ场景图像)，融合后的图像处理公式按照Mask*(α*ρ抠图图像+β*ρ抠图图像)+（1-Mask）*ρ场景图像。（所述公式中：Mask为图像掩膜，图像目标区域值为1，目标区域外值为0，ρ为密度值，α、β指系数）检测模型可对多场景下的手铐精准识别，同时将目标物的位置及所在X光图像信息记录标出。进一步的还可根据目标物位置信息推算目标物尺寸信息。
This dataset's intellectual property encompasses X-ray security inspection images of handcuffs captured from multiple angles and scenarios. Through processing procedures including image annotation, matting, enhancement, and fusion, it can serve as high-quality samples for training intelligent handcuff detection algorithm models, enabling accurate recognition of handcuffs hidden in other items or disguised as other objects. The generated intelligent detection model can be applied to various security inspection scenarios. 1. Data Source: X-ray sources are used to transmit handcuffs under multiple angles and scenarios, and the collected original X-ray image atlas database is established. 2. In-depth Data Processing: Pre-label the coordinate positions and item categories of the collected original X-ray images, and conduct matting processing on the handcuff images. Subsequently, perform augmentation processing such as geometric transformation and pixel transformation on the matted images and multi-scenario images respectively. 3. Rules for Generating Detection Model Algorithms: Fuse the processed handcuff images and scenario images according to the region matching principle based on density statistics (where pixel values represent physical density values). The fused region mask is used as the data label, and the fused image is used as deep learning sample data. Additionally, different average density differences can be obtained by adjusting the position of the matting area within the scenario image area, so as to train an intelligent detection model that can accurately locate and recognize handcuffs. The region matching principle follows the formula: Mask*(α*ρ_matting_image + β*ρ_scenario_image). The post-fusion image processing formula follows: Mask*(α*ρ_matting_image + β*ρ_matting_image) + (1-Mask)*ρ_scenario_image. (In the above formula: Mask is the image mask, with a value of 1 within the target area of the image and 0 outside the target area; ρ represents the density value, and α and β are coefficients.) The detection model can accurately recognize handcuffs in multiple scenarios, while recording and marking the position of the target object and the information of the X-ray image where it is located. Furthermore, the size information of the target object can be inferred based on its position information.
