当前位置：首页 > news >正文

购买服务器后如何做网站单页网站cms

news 2026/4/26 11:17:20

购买服务器后如何做网站,单页网站cms,校园网站建设培训的心得体会,wordpress没有描述本文是关于基于YOLOv4开发构建目标检测模型的超详细实战教程#xff0c;超详细实战教程相关的博文在前文有相应的系列#xff0c;感兴趣的话可以自行移步阅读即可#xff1a;《基于yolov7开发实践实例分割模型超详细教程》《YOLOv7基于自己的数据集从零构建模型完整训练、…本文是关于基于YOLOv4开发构建目标检测模型的超详细实战教程超详细实战教程相关的博文在前文有相应的系列感兴趣的话可以自行移步阅读即可《基于yolov7开发实践实例分割模型超详细教程》《YOLOv7基于自己的数据集从零构建模型完整训练、推理计算超详细教程》《DETR (DEtection TRansformer)基于自建数据集开发构建目标检测模型超详细教程》《基于yolov5-v7.0开发实践实例分割模型超详细教程》《轻量级模型YOLOv5-Lite基于自己的数据集【焊接质量检测】从零构建模型超详细教程》《轻量级模型NanoDet基于自己的数据集【接打电话检测】从零构建模型超详细教程》《基于YOLOv5-v6.2全新版本模型构建自己的图像识别模型超详细教程》《基于自建数据集【海底生物检测】使用YOLOv5-v6.1/2版本构建目标检测模型超详细教程》《超轻量级目标检测模型Yolo-FastestV2基于自建数据集【手写汉字检测】构建模型训练、推理完整流程超详细教程》《基于YOLOv8开发构建目标检测模型超详细教程【以焊缝质量检测数据场景为例】》最早期接触v3和v4的时候印象中模型的训练方式都是基于Darknet框架开发构建的模型都是通过cfg文件进行配置的从v5开始才全面转向了PyTorch形式的项目延续到了现在。 yolov4.cfg如下 [net] batch64 subdivisions8 # Training #width512 #height512 width608 height608 channels3 momentum0.949 decay0.0005 angle0 saturation 1.5 exposure 1.5 hue.1learning_rate0.0013 burn_in1000 max_batches 500500 policysteps steps400000,450000 scales.1,.1#cutmix1 mosaic1#:104x104 54:52x52 85:26x26 104:13x13 for 416[convolutional] batch_normalize1 filters32 size3 stride1 pad1 activationmish# Downsample[convolutional] batch_normalize1 filters64 size3 stride2 pad1 activationmish[convolutional] batch_normalize1 filters64 size1 stride1 pad1 activationmish[route] layers -2[convolutional] batch_normalize1 filters64 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters32 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters64 size3 stride1 pad1 activationmish[shortcut] from-3 activationlinear[convolutional] batch_normalize1 filters64 size1 stride1 pad1 activationmish[route] layers -1,-7[convolutional] batch_normalize1 filters64 size1 stride1 pad1 activationmish# Downsample[convolutional] batch_normalize1 filters128 size3 stride2 pad1 activationmish[convolutional] batch_normalize1 filters64 size1 stride1 pad1 activationmish[route] layers -2[convolutional] batch_normalize1 filters64 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters64 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters64 size3 stride1 pad1 activationmish[shortcut] from-3 activationlinear[convolutional] batch_normalize1 filters64 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters64 size3 stride1 pad1 activationmish[shortcut] from-3 activationlinear[convolutional] batch_normalize1 filters64 size1 stride1 pad1 activationmish[route] layers -1,-10[convolutional] batch_normalize1 filters128 size1 stride1 pad1 activationmish# Downsample[convolutional] batch_normalize1 filters256 size3 stride2 pad1 activationmish[convolutional] batch_normalize1 filters128 size1 stride1 pad1 activationmish[route] layers -2[convolutional] batch_normalize1 filters128 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters128 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters128 size3 stride1 pad1 activationmish[shortcut] from-3 activationlinear[convolutional] batch_normalize1 filters128 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters128 size3 stride1 pad1 activationmish[shortcut] from-3 activationlinear[convolutional] batch_normalize1 filters128 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters128 size3 stride1 pad1 activationmish[shortcut] from-3 activationlinear[convolutional] batch_normalize1 filters128 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters128 size3 stride1 pad1 activationmish[shortcut] from-3 activationlinear[convolutional] batch_normalize1 filters128 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters128 size3 stride1 pad1 activationmish[shortcut] from-3 activationlinear[convolutional] batch_normalize1 filters128 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters128 size3 stride1 pad1 activationmish[shortcut] from-3 activationlinear[convolutional] batch_normalize1 filters128 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters128 size3 stride1 pad1 activationmish[shortcut] from-3 activationlinear[convolutional] batch_normalize1 filters128 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters128 size3 stride1 pad1 activationmish[shortcut] from-3 activationlinear[convolutional] batch_normalize1 filters128 size1 stride1 pad1 activationmish[route] layers -1,-28[convolutional] batch_normalize1 filters256 size1 stride1 pad1 activationmish# Downsample[convolutional] batch_normalize1 filters512 size3 stride2 pad1 activationmish[convolutional] batch_normalize1 filters256 size1 stride1 pad1 activationmish[route] layers -2[convolutional] batch_normalize1 filters256 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters256 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters256 size3 stride1 pad1 activationmish[shortcut] from-3 activationlinear[convolutional] batch_normalize1 filters256 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters256 size3 stride1 pad1 activationmish[shortcut] from-3 activationlinear[convolutional] batch_normalize1 filters256 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters256 size3 stride1 pad1 activationmish[shortcut] from-3 activationlinear[convolutional] batch_normalize1 filters256 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters256 size3 stride1 pad1 activationmish[shortcut] from-3 activationlinear[convolutional] batch_normalize1 filters256 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters256 size3 stride1 pad1 activationmish[shortcut] from-3 activationlinear[convolutional] batch_normalize1 filters256 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters256 size3 stride1 pad1 activationmish[shortcut] from-3 activationlinear[convolutional] batch_normalize1 filters256 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters256 size3 stride1 pad1 activationmish[shortcut] from-3 activationlinear[convolutional] batch_normalize1 filters256 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters256 size3 stride1 pad1 activationmish[shortcut] from-3 activationlinear[convolutional] batch_normalize1 filters256 size1 stride1 pad1 activationmish[route] layers -1,-28[convolutional] batch_normalize1 filters512 size1 stride1 pad1 activationmish# Downsample[convolutional] batch_normalize1 filters1024 size3 stride2 pad1 activationmish[convolutional] batch_normalize1 filters512 size1 stride1 pad1 activationmish[route] layers -2[convolutional] batch_normalize1 filters512 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters512 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters512 size3 stride1 pad1 activationmish[shortcut] from-3 activationlinear[convolutional] batch_normalize1 filters512 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters512 size3 stride1 pad1 activationmish[shortcut] from-3 activationlinear[convolutional] batch_normalize1 filters512 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters512 size3 stride1 pad1 activationmish[shortcut] from-3 activationlinear[convolutional] batch_normalize1 filters512 size1 stride1 pad1 activationmish[convolutional] batch_normalize1 filters512 size3 stride1 pad1 activationmish[shortcut] from-3 activationlinear[convolutional] batch_normalize1 filters512 size1 stride1 pad1 activationmish[route] layers -1,-16[convolutional] batch_normalize1 filters1024 size1 stride1 pad1 activationmish##########################[convolutional] batch_normalize1 filters512 size1 stride1 pad1 activationleaky[convolutional] batch_normalize1 size3 stride1 pad1 filters1024 activationleaky[convolutional] batch_normalize1 filters512 size1 stride1 pad1 activationleaky### SPP ### [maxpool] stride1 size5[route] layers-2[maxpool] stride1 size9[route] layers-4[maxpool] stride1 size13[route] layers-1,-3,-5,-6 ### End SPP ###[convolutional] batch_normalize1 filters512 size1 stride1 pad1 activationleaky[convolutional] batch_normalize1 size3 stride1 pad1 filters1024 activationleaky[convolutional] batch_normalize1 filters512 size1 stride1 pad1 activationleaky[convolutional] batch_normalize1 filters256 size1 stride1 pad1 activationleaky[upsample] stride2[route] layers 85[convolutional] batch_normalize1 filters256 size1 stride1 pad1 activationleaky[route] layers -1, -3[convolutional] batch_normalize1 filters256 size1 stride1 pad1 activationleaky[convolutional] batch_normalize1 size3 stride1 pad1 filters512 activationleaky[convolutional] batch_normalize1 filters256 size1 stride1 pad1 activationleaky[convolutional] batch_normalize1 size3 stride1 pad1 filters512 activationleaky[convolutional] batch_normalize1 filters256 size1 stride1 pad1 activationleaky[convolutional] batch_normalize1 filters128 size1 stride1 pad1 activationleaky[upsample] stride2[route] layers 54[convolutional] batch_normalize1 filters128 size1 stride1 pad1 activationleaky[route] layers -1, -3[convolutional] batch_normalize1 filters128 size1 stride1 pad1 activationleaky[convolutional] batch_normalize1 size3 stride1 pad1 filters256 activationleaky[convolutional] batch_normalize1 filters128 size1 stride1 pad1 activationleaky[convolutional] batch_normalize1 size3 stride1 pad1 filters256 activationleaky[convolutional] batch_normalize1 filters128 size1 stride1 pad1 activationleaky##########################[convolutional] batch_normalize1 size3 stride1 pad1 filters256 activationleaky[convolutional] size1 stride1 pad1 filters255 activationlinear[yolo] mask 0,1,2 anchors 12, 16, 19, 36, 40, 28, 36, 75, 76, 55, 72, 146, 142, 110, 192, 243, 459, 401 classes80 num9 jitter.3 ignore_thresh .7 truth_thresh 1 scale_x_y 1.2 iou_thresh0.213 cls_normalizer1.0 iou_normalizer0.07 iou_lossciou nms_kindgreedynms beta_nms0.6 max_delta5[route] layers -4[convolutional] batch_normalize1 size3 stride2 pad1 filters256 activationleaky[route] layers -1, -16[convolutional] batch_normalize1 filters256 size1 stride1 pad1 activationleaky[convolutional] batch_normalize1 size3 stride1 pad1 filters512 activationleaky[convolutional] batch_normalize1 filters256 size1 stride1 pad1 activationleaky[convolutional] batch_normalize1 size3 stride1 pad1 filters512 activationleaky[convolutional] batch_normalize1 filters256 size1 stride1 pad1 activationleaky[convolutional] batch_normalize1 size3 stride1 pad1 filters512 activationleaky[convolutional] size1 stride1 pad1 filters255 activationlinear[yolo] mask 3,4,5 anchors 12, 16, 19, 36, 40, 28, 36, 75, 76, 55, 72, 146, 142, 110, 192, 243, 459, 401 classes80 num9 jitter.3 ignore_thresh .7 truth_thresh 1 scale_x_y 1.1 iou_thresh0.213 cls_normalizer1.0 iou_normalizer0.07 iou_lossciou nms_kindgreedynms beta_nms0.6 max_delta5[route] layers -4[convolutional] batch_normalize1 size3 stride2 pad1 filters512 activationleaky[route] layers -1, -37[convolutional] batch_normalize1 filters512 size1 stride1 pad1 activationleaky[convolutional] batch_normalize1 size3 stride1 pad1 filters1024 activationleaky[convolutional] batch_normalize1 filters512 size1 stride1 pad1 activationleaky[convolutional] batch_normalize1 size3 stride1 pad1 filters1024 activationleaky[convolutional] batch_normalize1 filters512 size1 stride1 pad1 activationleaky[convolutional] batch_normalize1 size3 stride1 pad1 filters1024 activationleaky[convolutional] size1 stride1 pad1 filters255 activationlinear[yolo] mask 6,7,8 anchors 12, 16, 19, 36, 40, 28, 36, 75, 76, 55, 72, 146, 142, 110, 192, 243, 459, 401 classes80 num9 jitter.3 ignore_thresh .7 truth_thresh 1 random1 scale_x_y 1.05 iou_thresh0.213 cls_normalizer1.0 iou_normalizer0.07 iou_lossciou nms_kindgreedynms beta_nms0.6 max_delta5 yolov4-tiny.cfg如下 [net] # Testing #batch1 #subdivisions1 # Training batch64 subdivisions1 width416 height416 channels3 momentum0.9 decay0.0005 angle0 saturation 1.5 exposure 1.5 hue.1learning_rate0.00261 burn_in1000 max_batches 500200 policysteps steps400000,450000 scales.1,.1[convolutional] batch_normalize1 filters32 size3 stride2 pad1 activationleaky[convolutional] batch_normalize1 filters64 size3 stride2 pad1 activationleaky[convolutional] batch_normalize1 filters64 size3 stride1 pad1 activationleaky[route] layers-1 groups2 group_id1[convolutional] batch_normalize1 filters32 size3 stride1 pad1 activationleaky[convolutional] batch_normalize1 filters32 size3 stride1 pad1 activationleaky[route] layers -1,-2[convolutional] batch_normalize1 filters64 size1 stride1 pad1 activationleaky[route] layers -6,-1[maxpool] size2 stride2[convolutional] batch_normalize1 filters128 size3 stride1 pad1 activationleaky[route] layers-1 groups2 group_id1[convolutional] batch_normalize1 filters64 size3 stride1 pad1 activationleaky[convolutional] batch_normalize1 filters64 size3 stride1 pad1 activationleaky[route] layers -1,-2[convolutional] batch_normalize1 filters128 size1 stride1 pad1 activationleaky[route] layers -6,-1[maxpool] size2 stride2[convolutional] batch_normalize1 filters256 size3 stride1 pad1 activationleaky[route] layers-1 groups2 group_id1[convolutional] batch_normalize1 filters128 size3 stride1 pad1 activationleaky[convolutional] batch_normalize1 filters128 size3 stride1 pad1 activationleaky[route] layers -1,-2[convolutional] batch_normalize1 filters256 size1 stride1 pad1 activationleaky[route] layers -6,-1[maxpool] size2 stride2[convolutional] batch_normalize1 filters512 size3 stride1 pad1 activationleaky##################################[convolutional] batch_normalize1 filters256 size1 stride1 pad1 activationleaky[convolutional] batch_normalize1 filters512 size3 stride1 pad1 activationleaky[convolutional] size1 stride1 pad1 filters255 activationlinear[yolo] mask 3,4,5 anchors 10,14, 23,27, 37,58, 81,82, 135,169, 344,319 classes80 num6 jitter.3 scale_x_y 1.05 cls_normalizer1.0 iou_normalizer0.07 iou_lossciou ignore_thresh .7 truth_thresh 1 random0 resize1.5 nms_kindgreedynms beta_nms0.6[route] layers -4[convolutional] batch_normalize1 filters128 size1 stride1 pad1 activationleaky[upsample] stride2[route] layers -1, 23[convolutional] batch_normalize1 filters256 size3 stride1 pad1 activationleaky[convolutional] size1 stride1 pad1 filters255 activationlinear[yolo] mask 1,2,3 anchors 10,14, 23,27, 37,58, 81,82, 135,169, 344,319 classes80 num6 jitter.3 scale_x_y 1.05 cls_normalizer1.0 iou_normalizer0.07 iou_lossciou ignore_thresh .7 truth_thresh 1 random0 resize1.5 nms_kindgreedynms beta_nms0.6 最开始的时候还是蛮喜欢这种形式的非常的简洁直接使用Darknet框架训练也很方便到后面随着模型改进各种组件的替换Darknet变得越发不适用了。YOLOv4的话感觉定位相比于v3和v5来说比较尴尬一些git里面搜索yolov4结果如下所示排名第一的项目是pytorch-YOLOv4地址在这里如下所示从说明里面来看这个只是一个minimal的实现官方的实现应该是仔细看的话会发现官方这里提供了YOLOv3风格的实现项目以及YOLOv5风格的实现项目本文主要是以YOLOv3风格的YOLOv4项目为基准来讲解完整的实践流程项目地址在这里如下所示首先下载所需要的项目如下下载到本地解压缩后如下所示网上直接百度下载这两个weights文件放在weights目录下如下所示然后随便复制过来一个自己之前yolov5项目的数据集放在当前项目目录下我是前面刚好基于yolov5做了钢铁缺陷检测项目数据集可以直接拿来用如果没有现成的数据集的话可以看我签名yolov5的超详细教程里面可以按照步骤自己创建数据集即可。如下所示这里我选择的是基于yolov4-tiny版本的模型来进行开发训练为的就是计算速度能够更快一些。修改train.py里面的内容如下所示 parser argparse.ArgumentParser() parser.add_argument(--weights, typestr, defaultweights/yolov4-tiny.weights, helpinitial weights path) parser.add_argument(--cfg, typestr, defaultcfg/yolov4-tiny.cfg, helpmodel.yaml path) parser.add_argument(--data, typestr, defaultdata/self.yaml, helpdata.yaml path) parser.add_argument(--hyp, typestr, defaultdata/hyp.scratch.yaml, helphyperparameters path) parser.add_argument(--epochs, typeint, default100) parser.add_argument(--batch-size, typeint, default8, helptotal batch size for all GPUs) parser.add_argument(--img-size, nargs, typeint, default[640, 640], help[train, test] image sizes) parser.add_argument(--rect, actionstore_true, helprectangular training) parser.add_argument(--resume, nargs?, constTrue, defaultFalse, helpresume most recent training) parser.add_argument(--nosave, actionstore_true, helponly save final checkpoint) parser.add_argument(--notest, actionstore_true, helponly test final epoch) parser.add_argument(--noautoanchor, actionstore_true, helpdisable autoanchor check) parser.add_argument(--evolve, actionstore_true, helpevolve hyperparameters) parser.add_argument(--bucket, typestr, default, helpgsutil bucket) parser.add_argument(--cache-images, actionstore_true, helpcache images for faster training) parser.add_argument(--image-weights, actionstore_true, helpuse weighted image selection for training) parser.add_argument(--device, default0, helpcuda device, i.e. 0 or 0,1,2,3 or cpu) parser.add_argument(--multi-scale, actionstore_true, helpvary img-size /- 50%%) parser.add_argument(--single-cls, actionstore_true, helptrain as single-class dataset) parser.add_argument(--adam, actionstore_true, helpuse torch.optim.Adam() optimizer) parser.add_argument(--sync-bn, actionstore_true, helpuse SyncBatchNorm, only available in DDP mode) parser.add_argument(--local_rank, typeint, default-1, helpDDP parameter, do not modify) parser.add_argument(--log-imgs, typeint, default16, helpnumber of images for WB logging, max 100) parser.add_argument(--workers, typeint, default8, helpmaximum number of dataloader workers) parser.add_argument(--project, defaultruns/train, helpsave to project/name) parser.add_argument(--name, defaultexp, helpsave to project/name) parser.add_argument(--exist-ok, actionstore_true, helpexisting project/name ok, do not increment) opt parser.parse_args() 终端直接执行 python train.py 即可。当然也可以选择基于参数指定的形式启动如下 python train.py --device 0 --batch-size 16 --img 640 640 --data self.yaml --cfg cfg/yolov4-tiny.cfg --weights weights/yolov4-tiny.weights --name yolov4-tiny 根据个人喜好来选择即可。启动训练终端输出如下所示训练完成截图如下所示训练完成我们来看下结果文件如下所示可以看到结果文件直观来看跟yolov5项目差距还是很大的评估指标只有一个PR图所以如果是做论文的话最好还是使用yolov5来做会好点。 PR曲线如下所示训练可视化如下所示 LABEL数据可视化如下所示 weights目录如下所示这个跟yolov5项目差异也是很大的yolov5项目只有两个pt文件一个是最优的一个是最新的但是yolov4项目居然产生了19个文件保存的可以说是非常详细了有点像yolov7但是比v7维度更多一些。感兴趣的话都可以按照我上面的教程步骤开发构建自己的目标检测模型。

http://www.hkea.cn/news/14420801/

相关文章：

南宁机关两学一做网站彩票网站开发定制

动态做网站网页点击量统计

cpa网站建设自己能做网站吗

怎么做盗号网站手机好看的旅游网站模版

用js做网站阅读量域名备案需要多久

51做网站建设企业官网公司logo效果图

营销网站报备怎样设置网站

常州网站定制wordpress新建页面显示数据库

建设公司网站需要准备哪些材料黑龙江新闻

用什么软件可以做网站宁城县建设局网站

常州个性化网站建设杭州市网站建设公司

用模板做企业网站做毕业设计的网站设计

网站策划的基本过程垂直行业门户网站

如何做网站赚一个服务器可以做多个网站吗

做一个网站平台的流程是什么绍兴网站建设优化

重庆网站建设的好处陕西省建设网官网陕西省建筑市场监督与诚信信息一体化平台

山西路桥建设集团有限公司网站wordpress汉化免费企业主题

网上做石材去哪个网站wordpress 批量导入用户名

青岛网站建设平台seo常规优化

深圳做分销网站设计平台推广引流

织梦小说网站源码asp网站后台失效

个人设计网站模板织梦技术个人网站模板下载

如何制作产品网站wordpress企业魔板

惠州惠阳网站建设福州建设企业网站

视频网站阿里云广州大学生网页设计大赛

一家公司多个网站备案怎么做网站广告代理商

创意品牌网站网站建设加推广需要多少钱

网站优化包括安卓内核级优化神器

辽宁省建设机械协会官方网站希尔顿酒店网络营销案例

指定图片做logo网站青岛团购网站建设