赛题说明: https://icver.blog.csdn.net/article/details/126121221?spm=1001.2014.3001.5502
retinanet数钢筋: https://github.com/spytensor/detect_steel_bar
YOLOv3数钢筋:
YOLOv3实战之钢筋数量AI识别: https://cloud.tencent.com/developer/article/1636776?from=article.detail.1636784
YOLOv3实战之钢筋智能识别改进方案: https://cloud.tencent.com/developer/article/1636784?from=article.detail.2046092
YOLOv5AI数钢筋 https://cloud.tencent.com/developer/article/2046092
缺点:
算法存在误判,将其他物体误判为钢筋头
重复检测,一个钢筋头被多个检测框标注
数据集: https://www.datafountain.cn/competitions/332/datasets
opencv钢管计数分析与方法比较
基于边缘与角度旋转的模板匹配-------Houg 霍夫圆检测
首先是加载图像,实现形态学梯度
然后使用霍夫变换,得到输出的圆数组,直接循环绘制颜色即可
参考链接: https://mp.weixin.qq.com/s/bNSnmeX2EAfKNB6vhtP_qQ