本文以石家庄市盲道为例，采用随机抽取方式，选取3个区的30条街道，对盲道使用情况进行实地调查。调查显示，69%现有盲道可以正常使用，其余存在着铺设不合理、被占用和破损情况，各占8%、11%、12%。本文从科学合理铺设、加强监管与维护、加强宣传教育三个方面提出改进意见，以改善盲道现状，提高使用率，促进视力障碍者参与社会生活，提高其生活质量。

In order to solve the problems of low measurement accuracy and poor algorithm adaptability in industrial measurement of part dimensions,a machine vision parts measurement system based on super-resolution model and improved Canny algorithm is designed and proposed.Firstly,the camera is calibrated to obtain the proportional relationship between the pixel distance and the actual size.Secondly,a super-resolution model is introduced and improved to pre-process the input image to obtain an image with clearer edge details.Thirdly,in order to solve the edge detection problem caused by poor lighting conditions in the industrial measurement environment,the Otsu algorithm and the double threshold segmentation algorithm are combined,and the Canny edge filtering algorithm is improved by using local threshold segmentation to obtain edge images.Finally,after simultaneously performing circle fitting and straight line fitting on the edge image,the part size results are marked based on the camera calibration results and super-resolution model parameters.The experiment results show that the measurement accuracy of this system can reach 0.001 mm,the average measurement error is 0.007 4 mm,and the overall measurement speed is 6.44 s,which can meet the needs of high-precision industrial measurement.

针对当前传统网络模型对交通标志识别精度低、检测不准确的问题，提出一种基于YOLOv8n优化、改进的Ghost-YOLOv8交通标志检测模型。使用GhostConv代替部分Conv，设计全新的C2fGhost模块代替部分C2f，减少了模型的参数量，提升了模型的检测性能；在Neck部分添加GAM注意力机制模块，强化特征中的语义信息和位置信息，提高了模型的特征融合能力；针对检测小目标时尺度不一导致语义信息的丢失，添加小目标检测层，增强深层语义信息与浅层语义信息的结合；使用GIoU边界损失函数代替原损失函数，提升了网络的边界框回归性能。实验结果表明，改进的模型在中国交通标志检测数据集TT100K中的精确度（Precision）及平均精度均值（mAP）相较于原模型分别提高了9.5、6.5个百分点，模型的参数量及模型大小相比原模型分别降低了0.223×109、0.2?MB。综合说明，该模型在减少模型参数量及大小的同时提高了检测精度，显著优于对比算法，也满足边缘计算设备的要求，具有实际的应用价值。

Aiming at the problems of low accuracy and inaccurate detection of traffic signs in the current traditional network model, a Ghost-YOLOv8 traffic sign detection model is proposed based on YOLOv8n optimization and improvement. First of all, using GhostConv instead of some Conv, designing a new C2fGhost module instead of some C2f, the model’s parameters is reduced and the detection performance of the model is enhanced. Secondly, a GAM attention mecha-<br>nism module is added to the Neck part to strengthen the semantic and positional information in the features, which improves the feature fusion ability of the model; then, for detecting the loss of semantic information when detecting small targets, it adds a small target detection layer to enhance the combination of deep semantic information and shallow semantic information. Finally, it uses the GIoU boundary border loss function to replace the original loss function, which improves the backbone of the network’s boundary frame. The experimental results show that the accuracy（Precision） and average accuracy average（mAP） of the improvement model in China traffic sign detection data set TT100K are increased by 9.5 and 6.5 percentage points compared with the original model. And the reduction of number of model parameters and model size are 0.223×109 and 0.2 MB, respectively. Comprehensive explanation, the model of this article improves the detection accuracy while reducing the amount and size of the model, which is significantly better than the comparison algorithm, and also meets the requirements of the edge computing equipment, and has practical application value.

针对无人机飞行高度高、目标尺度变化大、目标存在密集遮挡等问题,本文将对基于深度学习的无人机目标探测与识别以及视觉定位进行深入探讨,并设计出一套无人机目标探测与定位系统。研究选用Yolov5深度学习网络模型进行目标检测,基于Darknet深度学习框架完成端到端的训练预测,最后利用AprilTag视觉基准库完成无人机自身空间位置的辅助定位。测试结果表明,Yolov5模型的参数量只有5.3,准确率为97.41%,召回率为90.73%,mAP为83.2。AprilTag辅助定位的拟合精度达95%以上。研究设计的基于深度学习的智能化无人机视觉系统不仅具有实际的工程价值,更具有重要的社会意义。

To address the problems challenges such as high flight altitude, large changes in target scale, and the densely occluded targets, the study utilizes deep learning methods for target detection and identification and visual localization of UAV. A UAV target detection and positioning system is designed and discussed in depth. The study selects Yolov5 deep learning network model for target detection, completes end-to-end training prediction based on Darknet deep learning framework, and finally utilizes AprilTag visual benchmark library to complete the auxiliary localization of UAV's spatial positioning. The test results show that the parameter count of Yolov5 model is only 5.3. Meanwhile, the model achieves a precision of 97.41%, a recall rate of 90.73%, and an mAP of 83.2. The fitting precision of AprilTag assisted localization is more than 95%. The research and design of intelligent UAV vision system based on deep learning not only has actual engineering values, but also significant societal importance.