[train hyper-parameters: Namespace(batch_size=16, lr=0.001, lrf=0.01, num_classes=102)]
[epoch: 1]
train loss:0.2326 test accuracy:0.2873
[epoch: 2]
train loss:0.1724 test accuracy:0.4156
[epoch: 3]
train loss:0.1384 test accuracy:0.5061
[epoch: 4]
train loss:0.1115 test accuracy:0.5831
[epoch: 5]
train loss:0.0920 test accuracy:0.6381
[epoch: 6]
train loss:0.0767 test accuracy:0.6760
[epoch: 7]
train loss:0.0659 test accuracy:0.7628
[epoch: 8]
train loss:0.0550 test accuracy:0.7812
[epoch: 9]
train loss:0.0461 test accuracy:0.7616
[epoch: 10]
train loss:0.0401 test accuracy:0.8130
[epoch: 11]
train loss:0.0337 test accuracy:0.8154
[epoch: 12]
train loss:0.0281 test accuracy:0.8178
[epoch: 13]
train loss:0.0234 test accuracy:0.8606
[epoch: 14]
train loss:0.0214 test accuracy:0.8447
[epoch: 15]
train loss:0.0186 test accuracy:0.8594
[epoch: 16]
train loss:0.0155 test accuracy:0.8716
[epoch: 17]
train loss:0.0125 test accuracy:0.8680
[epoch: 18]
train loss:0.0101 test accuracy:0.8729
[epoch: 19]
train loss:0.0086 test accuracy:0.8680
[epoch: 20]
train loss:0.0069 test accuracy:0.8888
[epoch: 21]
train loss:0.0065 test accuracy:0.8814
[epoch: 22]
train loss:0.0047 test accuracy:0.9022
[epoch: 23]
train loss:0.0040 test accuracy:0.8888
[epoch: 24]
train loss:0.0036 test accuracy:0.8924
[epoch: 25]
train loss:0.0031 test accuracy:0.8949
[epoch: 26]
train loss:0.0032 test accuracy:0.8839
[epoch: 27]
train loss:0.0026 test accuracy:0.8998
[epoch: 28]
train loss:0.0020 test accuracy:0.9083
[epoch: 29]
train loss:0.0019 test accuracy:0.9083
[epoch: 30]
train loss:0.0016 test accuracy:0.9034
[epoch: 31]
train loss:0.0015 test accuracy:0.9046
[epoch: 32]
train loss:0.0016 test accuracy:0.9071
[epoch: 33]
train loss:0.0013 test accuracy:0.9144
[epoch: 34]
train loss:0.0011 test accuracy:0.9095
[epoch: 35]
train loss:0.0012 test accuracy:0.9108
[epoch: 36]
train loss:0.0011 test accuracy:0.9108
[epoch: 37]
train loss:0.0011 test accuracy:0.9059
[epoch: 38]
train loss:0.0011 test accuracy:0.9144
[epoch: 39]
train loss:0.0010 test accuracy:0.9108
[epoch: 40]
train loss:0.0009 test accuracy:0.9059
[epoch: 41]
train loss:0.0010 test accuracy:0.9205
[epoch: 42]
train loss:0.0008 test accuracy:0.9132
[epoch: 43]
train loss:0.0008 test accuracy:0.9120
[epoch: 44]
train loss:0.0009 test accuracy:0.9144
[epoch: 45]
train loss:0.0008 test accuracy:0.9120
[epoch: 46]
train loss:0.0008 test accuracy:0.9218
[epoch: 47]
train loss:0.0008 test accuracy:0.9181
[epoch: 48]
train loss:0.0007 test accuracy:0.9144
[epoch: 49]
train loss:0.0008 test accuracy:0.9083
[epoch: 50]
train loss:0.0007 test accuracy:0.9083
没有合适的资源?快使用搜索试试~ 我知道了~
温馨提示
基于DenseNet169网络对常见花类数据集的迁移学习项目,包含代码、数据集和训练好的权重文件,可直接运行。 项目总大小:394MB 本数据集分为以下102类:columbine、cyclamen、english marigold等等共102类别(每个类别均有40-200+张图片) 下载解压后的图像目录:训练集(6552张图片)、和测试集(818张图片) data-train 训练集-每个子文件夹放同类别的图像,文件夹名为分类类别 data-test 测试集-每个子文件夹放同类别的图像,文件夹名为分类类别 网络训练的时候采用cos 学习率自动衰减,训练了50个epoch。模型在测试集最好的表现达到92%精度。在run_results 目录下存有最好的权重文件,以及训练日志和loss、精度曲线等等 预测的时候,只需要运行predict即可,代码会自动将inference下所有图片推理,并取前三个概率最大类别的绘制在左上角 如果想要训练自己的数据集,请查看README文件
资源推荐
资源详情
资源评论
收起资源包目录
图像分类实战项目:基于DenseNet169网络对常见102鲜花数据集的迁移学习项目 (2000个子文件)
image_05352_result.jpg 99KB
image_05280_result.jpg 95KB
image_03918.jpg 92KB
image_05770.jpg 83KB
image_03537.jpg 83KB
image_07084.jpg 83KB
image_05752.jpg 83KB
image_03987.jpg 83KB
image_06346_result.jpg 82KB
image_03483.jpg 82KB
image_07193.jpg 78KB
image_07085.jpg 78KB
image_03476.jpg 78KB
image_06472.jpg 76KB
image_07712.jpg 76KB
image_03104.jpg 75KB
image_03486.jpg 75KB
image_05626.jpg 75KB
image_07071.jpg 74KB
image_07802.jpg 74KB
image_03701.jpg 72KB
image_07062.jpg 72KB
image_03514.jpg 72KB
image_05760.jpg 71KB
image_03990.jpg 70KB
image_07164.jpg 70KB
image_03536.jpg 70KB
image_05910.jpg 70KB
image_03505.jpg 69KB
image_03657.jpg 69KB
image_03515.jpg 69KB
image_03539.jpg 69KB
image_03114.jpg 69KB
image_06737.jpg 69KB
image_05777.jpg 68KB
image_07747.jpg 68KB
image_02511.jpg 68KB
image_03543.jpg 67KB
image_03658.jpg 66KB
image_01376.jpg 66KB
image_07826.jpg 66KB
image_03914.jpg 66KB
image_05943.jpg 65KB
image_07072.jpg 65KB
image_05764.jpg 65KB
image_03978.jpg 65KB
image_05792.jpg 65KB
image_03944.jpg 65KB
image_05779.jpg 65KB
image_05585.jpg 64KB
image_05748.jpg 64KB
image_07061.jpg 64KB
image_05789.jpg 64KB
image_00680.jpg 63KB
image_07049.jpg 63KB
image_05955.jpg 63KB
image_03496.jpg 63KB
image_03504.jpg 63KB
image_08019.jpg 63KB
image_05747.jpg 63KB
image_03659.jpg 63KB
image_03953.jpg 63KB
image_05589.jpg 62KB
image_07763.jpg 62KB
image_00649.jpg 62KB
image_07081.jpg 62KB
image_07073.jpg 62KB
image_00629.jpg 62KB
image_03676.jpg 62KB
image_07780.jpg 62KB
image_05774.jpg 62KB
image_05791.jpg 62KB
image_03661.jpg 62KB
image_07047.jpg 62KB
image_03473.jpg 62KB
image_01027.jpg 62KB
image_03687.jpg 62KB
image_07616.jpg 62KB
image_03699.jpg 61KB
image_03474.jpg 61KB
image_05604.jpg 61KB
image_07806.jpg 61KB
image_05621.jpg 61KB
image_05758.jpg 61KB
image_03500.jpg 61KB
image_01948.jpg 61KB
image_00959.jpg 61KB
image_01099.jpg 61KB
image_06200.jpg 60KB
image_04198.jpg 60KB
image_02918.jpg 60KB
image_07074.jpg 60KB
image_03700.jpg 60KB
image_05782.jpg 59KB
image_03503.jpg 59KB
image_05588.jpg 59KB
image_05661.jpg 59KB
image_01484.jpg 59KB
image_03529.jpg 59KB
image_03715.jpg 59KB
共 2000 条
- 1
- 2
- 3
- 4
- 5
- 6
- 20
资源评论
Ai医学图像分割
- 粉丝: 2w+
- 资源: 2128
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
最新资源
- (源码)基于Arduino和Nextion的HMI人机界面系统.zip
- (源码)基于 JavaFX 和 MySQL 的影院管理系统.zip
- (源码)基于EAV模型的动态广告位系统.zip
- (源码)基于Qt的长沙地铁换乘系统.zip
- (源码)基于ESP32和DM02A模块的智能照明系统.zip
- (源码)基于.NET Core和Entity Framework Core的学校管理系统.zip
- (源码)基于C#的WiFi签到管理系统.zip
- (源码)基于WPF和MVVM框架的LikeYou.WAWA管理系统.zip
- (源码)基于C#的邮件管理系统.zip
- 【yan照门】chen冠希(1323张) [2月25日凌晨新增容祖儿全94张].rar.torrent
资源上传下载、课程学习等过程中有任何疑问或建议,欢迎提出宝贵意见哦~我们会及时处理!
点击此处反馈
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功