CangWang

please set your gradle version ,my test project use these version in build.gradle classpath 'com.android.tools.build:gradle:3.0.1' classpath "com.android.tools.build:gradle-experimental:0.11.1" classpath "org.jetbrains.kotlin:kotlin-gradle-plugin:1.2.0" Can you tell me your versions or you can change your...

please check your Cmake has downloaded already,it will use cmake file to compile this project

please check your Cmake has downloaded already,it will use cmake file to compile this project

use this to Fix folder = os.path.join(data_dir, self.photo) files = os.listdir(folder) self.image_files[self.photo] = [os.path.join(folder, fi) for fi in files] for opt in [self.style, self.smooth]: files = os.listdir(opt) self.image_files[opt] = [os.path.join(opt,...

You have use how many epoch to change？ Use tune hyperparameters is another function to change ，and we can‘’t feel the source image and fake images direct。

[2020-10-30 00:40:02,215]-[train.py line:212]: === Epoch:[119/120],step:[310/312],img_size:[416],total_loss:20.4310|loss_ciou:4.3372|loss_conf:7.3171|loss_cls:8.7767|lr:0.0001 这是最后一个epoch的loss值

 可以看到所有的mAp都是偏低

是重新训练的，并未使用预训练模型

你loss是到达多少后无法下降，我总loss到达了20

> >  > > 使用mobilenetv3-Pytorch训练和只使用voc2007数据集，中途不检查mAp，训练了120次后，最后检查mAp，发现mAp只有0.178%是哪个步骤出错了吗？ > > 请问你的训练集和验证集是怎么划分的  划分是使用你的，我直接注释掉VOC2012的部分