我的名字呀 comments

Results 6 comments of


                                            我的名字呀

不加载预训练模型，从头开始训练

> 如果不加载预训练模型，从头开始训练，模型收敛效果好像差很多，且存在严重过拟合，有没有好的办法解决，求大神指点！！！ > 如果不加载预训练模型，从头开始训练，模型收敛效果好像差很多，且存在严重过拟合，有没有好的办法解决，求大神指点！！！你好，如何不加载预训练模型开始训练，好奇为什么会过拟合

When I try to run generate_blur.py， I got this RuntimeError

I download the pre_trained model vgg19-dcbb9e9d.pth in C:\, I can run generate_bule.py and also get the following print : Sample #0/10 Sample #1/10 Sample #2/10 Sample #3/10 Sample #4/10 Sample...

some error in code

> > Recently I do some experiment with this code,but find some error in your code.The first error is "IndexError: The shape of the mask [32, 8732] at index 0...

Help for custom ssd training

The same question with me. And I'm confused with the Recall and Precision, I run the eval.py then get 80% Recall but 3% Precison, save me...

precision is too low,who know why.thx

I got the same problem, save me.

ValueError: non-broadcastable output operand with shape (11725,) doesn't match the broadcast shape (1,11725)

I trained custom dataset, and I got this problem, too. Have you work it out?